Skip to main content

Full text of "High frequency properties of plasma"

See other formats


NASA TECHNICAL 

TRANS L ATI O N 



I 







I NASA TT F-449 



(! 



/ 













m 




^^^^s 


o 




^^^1^ 


ac 












1- 




I^^^^^H 


□ 


^"^^ 


S 


o 


■M^^ 


0" 




J3 


^^^ 


a 




UJ 


S^^SS 


X" 




= 


s 




^^^i^ 


a 




^^^^■" 


z 










3 







LOAN COPY: RETL 
AFWL {WLIL-2) 
KIRTLAND AFB, H MEX 



HIGH FREQUENCY 
PROPERTIES OF PLASMA 

K. D. SinePnikov, Editor-in-Chief 

Academy of Sciences, Ukrainian SSR, 
Izdatel'stvo "Naukova Gumka," 
Kiev, 1965 



NATIONAL AERONAUTICS AND SPACE ADMINISTRATION • WASHINGTON, D. C. # MARCH 1967 



TECH LIBRARY KAFB, NM 



QQb^D3M 
NASA TT F-449 



HIGH FREQUENCY PROPERTIES OF PLASMA 
K. D. Sinernikov, Editor-in-Chief 



Translation of "Vysokochastotnyye Svoystva Plazmy. 
Academy of Sciences, Ukrainian SSR, 
Izdatel'stvo "Naukova Dumka," 
Kiev, 1965. 



NATIONAL AERONAUTICS AND SPACE ADMINISTRATION 



For sale by the Clearinghouse for Federal Scientific and Technical Information 
Springfield, Virginia 22151 - Price $3.75 



TABLE OF CONTENTS 



HIGH FREQUENCY PROPERTIES OF PLASMA 

Page 
SECTION I: HIGH jFREQU ENCY PLASMA HEATING 1 

INVESTIGATION OF THE ENERGY OF CHARGED PARTICLES EMANATING 
FROM A MAGNETIC TRAP DURING HIGH FREQUENCY HEATING 

N. I. Nazarov, A. I. Yermakov, V. T. Tolok 1 

MEASUREMENT OF THE PERPENDICULAR ENERGY COMPONENT AND THE 
PLASMA DECAY TIME DURING HIGH FREQUENCY HEATING 

N. I. Nazarov, A. I. Yermakov, V. T. Tolok 6 

HIGH FREQUENCY ENERGY ABSORPTION BY A PLASMA IN ION 
CYCLOTRON RESONANCE IN STRONG, HIGH FREQUENCY FIELDS 

V. V. Chechkin, M. P. Vasil'y^v, L. I. Grigor'yeva, 

B . I . Smerdov 11 

INVESTIGATION OF CONDITIONS PRODUCING A DENSE PLASMA IN 
A METALLIC CHAMBER AND ITS HIGH FREQUENCY HEATING 

0. M. Shvets, S. S. Ovchinnikov, V. P. Tarasenko, 

L. V. Brzhechko, 0. S. Pavlichenko, V. T. Tolok 22 

HIGH FREQUENCY PLASMA HEATING 

K. N, Stepanov 33 

DIELECTRIC CONSTANT OF A PLASMA IN A DIRECT PINCH MAGNETIC 
FIELD AND IN A DIRECT HELICAL MAGNETIC FIELD 

V. F. Aleksin, V. I. Yashin 55 

SECTION II: LINEAR PLASMA OSCILLATIONS 64 

KINETIC THEORY OF ELECTROMAGNETIC WAVES IN A CONFINED PLASMA 

A. N. Kondratenko 64 



iil 



Page 

KINETIC THEORY OF A SURFACE WAVE IN A PLASMA WAVE GUIDE 

M. F. Gorbatenko, V. I. Kurllko 75 

SINGULARITIES OF AN ELECTROMAGNETIC FIELD IN A NONUNIFORM, 
MAGNETOACTIVE PLASMA 

V. V. Dolgopolov 80 

EXCITATION OP A MAGNETOHYDRODYNAMIC WAVE GUIDE IN A 
COAXIAL LINE 

S. S. Kalmykova, V. I. Kurllko 89 

THEORY OF MAGNETOHYDRODYNAMIC WAVE SCATTERING AT THE END 
OF A WAVE GUIDE 

V. I. Kurllko 91 

DETERMINATION OF PLASMA TEMPERATURE AND DENSITY DISTRIBUTION 
BY REFRACTION AND DAMPING OF A BEAM 

V. L. Sizonenko, K. N. Stepanov 100 



SECTION III: PLASMA NONLINEAR OSCILLATIONS AND WAVE INTERACTION 110 

EXCHANGE OF ENERGY BETWEEN HIGH FREQUENCY AND LOW FREQUENCY 
OSCILLATIONS IN A PLASMA 

V. D. Fedorchenko , V. I. Muratov, B. N. Rutkevich 110 

DISSIPATION OF PLASMA OSCILLATIONS EXCITED IN A CURRENT- 
CARRYING PLASMA 

Ye. A. Sukhomlln, V. A. Suprunenko, N. I. Reva, 

V. T. Tolok 119 

DAMPING OF INITIAL PERTURBATION AND STEADY FLUCTUATIONS 
IN A COLLISIONLESS PLASMA 

A. I. Akhlyezer, I. A. Akhiyezer, R. V. Polovln 127 

CHARGED PARTICLE INTERACTION WITH A TURBULENT PLASMA 

I . A. Akhlyezer 133 



IV 



Page 

THEORY OF NONLINEAR MOTIONS OF A NONEQUILIBRIUM PLASMA 

I . A. Akhlyezer 136 

NONLINEAR PROCESSES IN A UNIFORM AND ONE-COMPONENT PLASMA 

N. A. Khizhnyak, A. M. Korsunskiy 143 

INDUCED SCATTERING OF LANGMUIR OSCILLATIONS IN A PLASMA 
LOCATED IN A STRONG MAGNETIC FIELD 

V. D. Shapiro, V. I, Shevchenko 151 

NONLINEAR THEORY OF LOW FREQUENCY OSCILLATIONS EXCITED 
BY AN ION BUNDLE IN A PLASMA 

D . G. Lomlnadze , V. I . Shevchenko 159 

NONLINEAR PHENOMENA IN A PLASMA WAVE GUIDE (ION 
CYCLOTRON RESONANCE AT A DIFFERENCE FREQUENCY) 

B. I. Ivanov 172 



SECTION IV: EXCITATION OF PLASMA OSCILLATIONS 180 

RADIATION OF ELECTRONS IN THE PLASMA-MAGNETIC FIELD 
BOUNDARY LAYER 

V. V. Dolgopolov, V. I. Pakhomov, K. N. Stepanov 180 

RADIATION OF LOW FREQUENCY WAVES BY IONS AND ELECTRONS 
OF A NON-ISOTHEEMIC MAGNETOACTIVE PLASMA 

V. I. Pakhomov 183 

BRAKING OF RELATIVISTIC PARTICLES IN LOW ATMOSPHERIC 
LAYERS 

V. B. Krasovltskiy , V. I. Kurilko 199 

EXCITATION OF WAVES IN A CONFINED PLASMA BY MODULATED 
CURRENTS 

A. N. Kondratenko 203 



HIGH FREQUENCY PROPERTIES OF PLASMA 
PLASMA PHYSICS AND THE PROBLEMS OF CONTROLLED THERMONUCLEAR REACTION 



ABSTRACT 

These articles present the results derived from 
theoretical and experimental investigations of high 
frequency plasma properties : the methods of high fre- 
quency plasma heating, propagation of electromagnetic 
waves in a magnetoactive plasma, thermal radiation of 
a plasma, and development of instabilities when employing 
high frequency methods of plasma heating. A description 
is provided of experimental equipment developed for high 
frequency heating and containment of a plasma. 

This collection is designed for scientific researchers 
and engineers dealing with the problems of a plasma and 
its technical application, as well as for students and 
graduate students in the physics departments of universi- 
ties and physical-technical institutes. 



SECTION I 
HIGH FREQUENCY PLASMA HEATING 



INVESTIGATION OF THE ENERGY OF CHARGED PARTICLES EMANATING 
FROM A MAGNETIC TRAP DURING HIGH FREQUENCY HEATING 

N. I. Nazarov, A. I. Yermakov, V. T. Tolok 

The methods of resonance excitation of the eigen fluctuations of a /5* 
plasma cylinder by outer electromagnetic fields have been extensively em- 
ployed in plasma heating with high frequency fields. Spatially periodic 
electromagnetic fields may be employed to excite the eigen fluctuations in 
a plasma located in a constant magnetic field to frequencies which are close 



* Note: Nimibers in the margin indicate pagination in the original foreign 
text. 



to the gyrofrequency of ions co « co . (an ion cyclotron wave) or to frequencies 
ojjj^ < 0) « (jOjjg (rapid magnetosound wave) , where m^^ is the electron gyrofre- 
quency and tOti- is the ion gyrofrequency. In the first case, the energy of 

electromagnetic fluctuations is transmitted directly to the ions, and in the 
second case the energy is transmitted to the plasma electrons. 

The articles (Ref. 1, 2) have investigated the conditions for the reso- 
nance excitation of these fluctuations, their propagation, and damping. It 
has been found that even at high electron temperatures the damping is signi- 
ficant and is caused by a collisionless mechanism. It was also found that, /6 
when this excitation method is employed, high frequency power is transmitted 
to the plasma very effectively. 

This study presents the results derived from measuring the energy of 
ions and electrons passing along the magnetic field, when ion cyclotron and 
rapid magnet ohydrodynamic waves are employed for heating the plasma. The 
experiment was performed on a "Sneg" apparatus, which has been described in 
great detail previously (Ref. 1). The eigen fluctuations were excited in the 
plasma by spatially periodic electromagnetic fields at a frequency of 10 Mc 
with the appropriate selection of the magnetic field strength Hg . In contrast 

to preceding experiments, the power of the high frequency (hf) generator was 
increased to 300 kw. 

In order to increase the power introduced into the plasma, the pulse of 
the hf generator was programmed so that the strong loading on the 
circuit at the moment of its resonance loading by the plasma was compensated 
by a corresponding voltage increase in the pulse from the hf generator (Figure 
1). Thus, the necessity of a special electrical strengthening of the hf cir- 
cuit was avoided, even when a power greater than 100 kw was introduced into 
the plasma. 

The energy of the charged particles was measured by a transit time electro- 
static analyzer (Ref. 3) and a multigrid probe (Ref. 4). The first method 
made it possible to study the energy spectrum and the mass composition of 
plasma ions; the second method made it possible to measure the energy of 
ions and electrons. The plasma electron temperature was determined by a 
spectral method. 

The input slit of the analyzer, which was located 25 cm behind a mag- 
netic mirror, cut out a narrow plasma flux, from which an ion bundle was 
separated after passing a separation device. The energy of the bundle 
was analyzed by the electric field of the flat condenser. Ion fluxes were 
recorded by an ion-electron converter, which changed the ion bundle into a 
bundle of electrons accelerated up to 20 kev, after deflection in the 
analyzing condenser. These electrons were detected by a plastic scintilla- 
tor with a photoelectric multiplier. 

In order to study the mass composition of the plasma ions, the flight 




Figure 1 

time was measured by ions having a drift section 56 cm long. When the analy- 
zer modulator was supplied with voltage having a rectangular form, it was 
possible to obtain short pulses of the ion current (t = 1; 0.5; 0.2 microseconds) 
Due to a difference in the velocities of different ions, the ions were separ- 
ated by mass in the drift space. The flight time of ions in the drift sec- /7 
tion was used to determine the ion velocities and masses, respectively. By 
measuring the amplitude of the current signals at different periods of time 
and by changing the voltage on the analyzing condenser, it was possible to 
record the energy spectra of ions having different masses and to observe 
the change in ion energy during heating, by means of this analyzer. 

In order to exclude the scatter of ion current pulses, the result was 
averaged over ten measurements (the ion energy was determined with the 
analyzer with an accuracy of 8%) . 

Figure 2 presents the oscillograms of a typical signal from the photo- 
multiplier, whose magnitude was proportional to the current of ions having 
an energy of 1500 ev — a; b — represents the suppression signal of micro- 
waves having a wavelength of 8 mm. 

Figure 3 shows the energy spectra of plasma protons (the coordinates: 
the distribution functions i|; — ion energy E.) when the plasma is heated by 

means of ion cyclotron waves for two voltages on the hf circuit (curve 1 — 
U(, = 28 kv; curve 2 — U = 32 kv) . The plasma density to be measured during 

heating was no less than 2*10^^ cm~^. As may be seen from the figure, the 
energy at the spectrvim maximum amounts to 2 kev. With an increase in the 
voltage Ug on the exciting coil of the hf circuit, the mean ion energy in 
the spectrum increases proportionally to the square of the voltage U^. How- 
ever, during the heating pulse it remains almost constant, which points to /8 
large losses which are apparently caused by overcharging. 

For purposes of comparison, the proton energy spectrum is presented which 
was recorded when the plasma was heated by a rapid magnetosound wave with a 



' ^-^,^ 




Figure 2 



voltage of 32 kv on the hf circuit (Figure 4) 
trum maximum is 150 ev In all in this case. 



The ion energy at the spec- 



As may be seen from the ion mass-spectrograms (Figure 5) when the plasma 
is heated by an ion cyclotron wave, there are only hydrogen ions HJ, H|', EJ 

in the plasma. The presence of these ions is apparently due to the fact that 
the plasma electron temperature is low (20-25 ev) . It is particularly inter- 
esting to note that all three types of hydrogen ions have approximately the /9 
same energy, although the resonance acceleration occurs only for H^. 

For purposes of control, the ion energy was measured by another method 
— multigrid probe with retarding potential (Ref. 4). This made it possible 
to record the energy spectrum of electrons. The multigrid probe was placed 
at a distance of 10 cm from the magnetic mirror. The plasma density was 
greatly reduced by means of a diaphragm having several 0.1 mm openings. It 
was possible to separate an electron bundle or an ion bundle, depending on 
the sign of the pulling voltage on the first grid; the bundles were analyzed 
by the retarding potential which was supplied to the second grid. In order 
to decrease the possibility of ionization, a differential pumping was employed 
to maintain a vacuum of 6.7-10-'* n/m^ within the probe walls. 



'f'. 


rel. 
unxt 




1 






15 




/ 


AJ 


2 


\ 


10 




// 


A 


5 

n 


/ 


y 







500 1000 tSOO 2000 2500 Ei.ev 



Figure 3 




100 200 300 £,js.v 
Figure 4 




Figure 5 



The results derived from measuring the Ion energy spectrtjm by means of 
the multi-grid probe confirmed the results obtained with analyzer measure- 
ments. The proton energy spectrum maximum was in the region of 2 kev when 
the plasma was heated with Ion cyclotron waves. The electron energy remained 
low, and amounted to 30 ev in all. This result also coincides closely with 
the results derived from measuring the electron energy by the spectral 
methods. 

As would be expected, the measurements of the energy spectrvnn of ions 
and electrons by the multigrid probe, when the plasma was heated with a 
rapid magnetosound wave, showed that the electron and ion energies were 
approximately the same (150 ev) . The ion energies obtained by measurements 
with the probe and the analyzer also coincided fairly well. 



REFERENCES 



Nazarov, N. I., Yermakov, A. I., Lobko, A. S. , Bondarev, V. A., Tolok, 
V. T., Sinel'nikov, K. D. Zhurnal Tekhnicheskoy Fiziki, 32, 5, 536, 
1962. 



2. Nazarov, N. I., Yermakov, A. I., Dolgopolov, V. V., Stepanov, K. N. , 

Tolok, V. T. Yadernyy Sintez, 3, 255, 1963. 

3. Kalmykov, A. A., Timofeyev, A. D. , Pankrat'yev, Yu. I., Tereshin, V. I., 

Vereshchagin, V. L. , Zlatopol'skiy, L. A. Pribory i Tekhnlka Ek- 
sperlmenta, 5, 142, 1963. 

4. Bulyginskly, D. G., Galaktinov, B. V., Dalmatov, K. A., Ovsyanikov, V.A. 

Zhurnal Tekhnicheskoy Fiziki, 33, 183, 1963. 



MEASUREMENT OF THE PERPENDICULAR ENERGY COMPONENT AND THE /lO 

PLASMA DECAY TIME DURING HIGH FREQUENCY HEATING 

N. I. Nazarov, A. I. Yermakov, V. T. Tolok 

The correct determination of the charged particle temperature is of 
paramount importance when studying the effectiveness of plasma heating. 
One convenient method of determining the plasma temperature consists of 
measuring its thermal diamagnetism by an external diamagnetic probe which 
includes a coliamn of the plasma to be heated. The method is based on 
measuring the difference between the strengths of the magnetic field out- 
side and within the plasma AH, which is a function of the gasokinetic 



(-^) 



pressure. For a plasma with small g j g = ^^ j , this difference is deter- 
mined by the expression 



AH = Ho - H^ithin = 



= itIE (1) 

Ho 



where p is the gasokinetic pressure. With a plasma having quasineutrality, 
we have 



p = nk {Tj_i + T^,), 



(2) 



where n is the plasma density; k is the Boltzmann constant. T,. and Tj^ 

are the perpendicular ion and electron temperatures, respectively. Thus, 
by measuring AH and knowing the plasma density, we may compute its tempera- 
ture. 

The temperature was measured by this method on a "Sneg" apparatus 




Figure 1 



(Ref. 1). Similar measurements were performed in (Ref. 2). The plasma 
was either produced by a powerful ion cyclotron wave (Ref. 1), or by a 
rapid magnetosound wave (Ref. 3). Hydrogen in the 0.13-0.4 n/m^ pressure 
range was used as the process gas. The waves were excited at a frequency 
of fg = 10^ cps, and the pulse high frequency power, transmitted to the 
plasma, was 100 kw. In order to avoid transitional processes related to 
the sharp change in plasma density during the initial period, a coupled, 
high frequency pulse, whose envelope is shown in Figure 1, was supplied to 
the exciting coil. The first pulse produced a plasma, and the second pulse 
was employed to heat it. The duration of the pulses, the amplitude, and 
the interval between them could be changed Independently over very wide 
Intervals. 



The period during which the strength of the pulse magnetic field Hg 
changed was 24 microseconds. The strength of the field was selected so 
that either the ion cyclotron wave, or the rapid magnetosound wave, was 
excited resonantly at the time the second high frequency pulse came into 
operation. The plasma density was measured by a microwave Interferometer 
at the wavelengths 8.2 and 4 mm. 



/ll 



The quantity AH, which was caused by the plasma dlamagnetism, was de- 
termined by measurements of the electromotive force (emf) by a dlamagnetlc 
probe. The probe consisted of two colls, one of which Included the plasma 
column. The other coll was employed to compensate for the emf caused by a 
change in the strength of the confining magnetic field Kg. In order to 
eliminate the emf produced by the propagated ion cyclotron wave, a five-unit 
low-frequency filter was employed, which Intersected all frequencies above 
3 Mc. The probe was located in the region of the "magnetic beach", at a 
distance of 30 cm from the edge of the exciting coil. In order to decrease 
the effect of attenuation of the dlamagnetlc signal, due to reflection of 
the charged particles from the walls, the diameter of the discharge chamber 
was increased up to 8 cm in the region of the dlamagnetlc probe, while the 




Figure 2 

diameter of the plasma with n > 10-^^ cm~^ equalled 3.5 cm- 
Figure 2 shows an oscillogram of the dlamagnetlc signal — a, obtained 
when the plasma was heated with an ion cyclotron wave; b — represents the 
Interferogram of an 8-mm signal. The gasokinetlc plasma pressure increased 
very rapidly (in~10 microseconds), and then barely changed until the end 
of the high frequency pulse. Since the plasma density changed very little /12 
— (1.2 - 1.5) X 10^^ cm~^ — during the high frequency pulse, and since 
it decreased slowly after it had ended (t ~ 270 microseconds) , it can be 
assumed that the dlamagnetlc signal was proportional to the rate at which 

the perpendicular energy component of the plasma -r— (Ti , + Tj^) changed. 

Figure 3, a shows an oscillogram of a dlamagnetlc signal. In order to deter- 
mine the plasma temperature, AH was calculated after integration of the dla- 
magnetlc signal. The integrated dlamagnetlc signal is shown in Figure 3, b. 
The total value of T, thus obtained for resonance excitation of an ion cyclo- 
tron wave, amounted to 1 kev. The electron temperature, determined by the 
spectral method, was in this case 20-30 ev. Thus, the ion temperature was 
measured indirectly. The plasma temperature was determined by this same 
method when it was heated by a rapid magnetosound wave. In this case 
T = 200-300 ev. The values of T, obtained according to the dlamagnetlc 
signal, closely coincide with the measured energy of charged particles em- 
Inatlng from the system along the magnetic field (Ref. 4). The small dl- /13 
vergence between T. and T,, is caused by the fact that measurements of the 

plasma temperature according to the dlamagnetlc signal give an average (over 
the plasma column cross section) temperature. 

Ho 
Figure 4 shows the dependence of the ion temperature T-j^ on (Hq — 

"ci 
the strength of the outer magnetic field, H . — the strength of the magnetic 

field at which the gyrof requency of a proton equals 10 Mc) . It can be seen 
that maximum heating occurs with resonance excitation of an ion cyclotron 
wave. 




Figure 3 



'..- 


kev 














IP 


















f \ 






(\7^ 










' \ 






as 

0^5 








1 




V 










J 




\ 








^ 




-. 


. 


I 



V 12 

Figure 4 






The ion temperature barely changes during heating (see Figure 3) . 
There is a very rapid temperature decrease after the high frequency pulse 
is recorded. This change in the ion temperature points to the presence 
of great losses. Since the ion temperature is considerably greater than 
the electron temperature when this method of plasma heating is employed, 
it may be assumed that one of the mechanisms for rapid ion cooling is their 
energy loss when colliding with electrons. However, for a plasma with 
Hg ~ 2*10^^ and Tg~ 25 ev, the time required to cool hot ions must be 100 
microseconds. In actuality, the cooling time equalled 10 microseconds. 
Therefore, there is no basis for assuming that this is the main loss 
mechanism. 



Another mechanism for rapid ion cooling may be their overcharging, 
since under the experimental conditions the highly-ionized plasma column 
was surrounded with a weakly-ionized cold plasma which was in contact with 
the discharge chamber walls. For a plasma with an electron density on the 



^T3 



order of 10^^ cm-^ and a neutral gas pressure at the chamber walls of 
approximately 1.33 "10"^ n/m^, the probability of overcharging exceeds the 
ionization probability. Consequently, the losses to overcharging may be 
considerable, and they continue to increase with an increase in the ion 



IS 

fO 


• . ' ' 

micros ec 










*% 


N 


1 






5 




















\ 





m 200 600 iOOO ZOOOfi^V 



Figure 5 



energy. Figure 5 shows the dependence of the ion cooling time on their 
energy. The solid line corresponds to the time required for overcharging 
the ions for a density of the surrounding gas of ng = 6* 10-^^ cm~^. This 
density was selected so that a comparison could be made between the value 
obtained theoretically and the experimentally measured value t-j, for 

T-j^ = 200 ev. The nature of the dependence of T-p. on T- shows that under 

experimental conditions the energy losses are primarily caused by ion 
overcharging . 

Thus, these experiments enable us to draw the conclusion that a /14 
plasma may be heated to a temperature exceeding 1 kev by means of resonance 
excitation of an ion cyclotron wave. The limiting value T^ is determined 
by the apparatus parameters. In addition, the results obtained provide a 
basis for assuming that — when a hot plasma is insulated from the chamber 
walls by "vacuum interstratif ication" — the time the plasma may be contained 
may be increased considerably, under the condition that there are no other 
loss mechanisms. 



REFERENCES 

Nazarov, N. I., Yermakov, A. I., Lobko, A. S. , Bondarev, V. A., Tolok, 
V. T., Sinel'nikov, K. D. Zhurnal Tekhnicheskoy Fiziki, 32, 5, 536, 
1962. 

Hooke, W. H., Rothman, M. A., Adam, I. Bull. Am. Phys. Soc. Ser. II, 8, 
174, 1963. 

Nazarov, N. I., Yermakov, A. I., Dolgopolov, V. V., Stepanov, K. N. , 
Tolok, V. T. Yadernyy Sintez, 3, 255, 1963. 



10 



4. Nazarov, N. I., Yermakov, A. I., Tolok, V. T. Present Collection, 5. 



HIGH FREQUENCY ENERGY ABSORPTION BY A PLASMA IN ION CYCLOTRON /15 
RESONANCE IN STRONG, HIGH FREQUENCY FIELDS 

V. V. Chechkin, M. P. Vasil'yev, L. I. Grlgor'yeva, 

B . I . Smerdov 

This article represents a continuation of studies we performed pre- 
viously (Ref . 1, 2) on high frequency energy absorption by a plasma in 
ion cyclotron resonance. 

As is well known (Ref. 3-5), the heating of a plasma by a variable 
field at a frequency which is close to the ion cyclotron frequency is very 
effective, if there are mechanisms leading to energy thermalization of the 
orderly motion of plasma particles in the field of an ion cyclotron wave. 
High frequency energy absorption by the plasma may occur, in particular, 
due to "close" collisions of ions which are in resonance with other types 
of ions, electrons, and neutral atoms in a cold plasma. This absorption 
may also be due to "collisionless" cyclotron damping which is caused by the 
thermal motion of ions in a high temperature plasma. In both cases, it is 
assumed that the amplitude of the high frequency field is fairly small, so 
that ions receiving energy in the wave field can transmit it to other parti- 
cles. If this condition is not fulfilled, nonlinear processes must arise 
in the plasma, due to which the ion distribution function is essentially 
distorted. Distortion of the ion distribution function by the ion cyclotron 
wave with a finite amplitude leads to a decrease in the wave absorption co- 
efficient down to a small value which equals, in order of magnitude, the 
absorption coefficient for pair collisions at a given temperature. 

The expression obtained in the case of a high-temperature plasma 
(Ref. 5) for a critical field strength of an ion cyclotron wave — which 
leads to a significant distortion of the ion distribution function when 
it is exceeded — can be written as follows 









((ot) 

2Tr 
where w is the wave frequency; k . = r — ; A — axial wave length; H — con- 

II A 

stant magnetic field; t — relaxation time of ions due to ion- ion colli- /16 
sions; N. — ion density; T^ — ion temperature. 

11 



In the case of a cold plasma, the critical strength of the wave 
electric field (at the absorption maximum) is as follows 

Ecr--^ l/i;. (2) 

^^eff V M 

When this value is reached, the velocity of the ordered motion of an ion 
liquid, with respect to electrons, equals the thermal ion velocity. M — 
ion mass; e — ion charge; Yeff — effective frequency of ion collisions. 

If the velocity of the relative motion of ions and electrons in the 

field of an ion cyclotron wave is greater than the ion thermal velocity, 

"bunched" instability may arise in the plasma, which is related to the excita- 

eH 
tion of high frequency (as compared with 'J^i=T7~) longitudinal fluctuations, 

vifcose increasing increment is considerably greater than the cyclotron ion frequency, 
and the wavelength is considerably less than the ion cyclotron wavelength. 
The excitation of these fluctuations by an ion bundle moving in a direction 
which is perpendicular to a constant magnetic field has been investigated in 
(Ref. 6-8). 

Similar small-scale electrostatic plasma fluctuations, which take place 
in a wide frequency and wave number range, must lead to an increased exchange 
of energy between plasma ions and electrons (as compared with the exchange 
caused only by Coulomb collisions), and must also lead to a significant in- 
crease in all the transfer coefficients across a constant magnetic field. 

Let us examine certain results of experimental studies in this light 
(Ref. 9, 10). In these studies, the field strength of an ion cyclotron 
wave exceeds by at least one order of magnitude the critical field strength 
(1) , and the time of ion-ion relaxation is either a comparable with the 
plasma decay time (Ref. 9) or considerably exceeds it (Ref. 10). For this 
reason, the strong high frequency energy absorption by the plasma and the 
ion heating, observed in these studies, cannot — in our opinion — be 
caused by cyclotron damping. In addition, the study (Ref. 11), which was 
carried out on the same apparatus as was employed in (Ref. 10), observed a 
rapid plasma decay if the high frequency power introduced into the plasma 
exceeded a certain critical value. This rapid decay was apparently caused /17 
by "bunched" instability. 

The phenomenon of anomalous plasma diffusion across a magnetic field, 
produced for a critical value of a high frequency field strength with a 
frequency close to ion cyclotron frequency, was discovered and studied in 
detail in (Ref. 2). In particular, this study found that increased diffu- 
sion occurs if the velocity, acquired by ions in an azimuthal high frequency 
field during the period between collisions, is comparable to the thermal ion 
velocity or exceeds it — i.e., if relationship (2) is fulfilled. 

The study (Ref. 12) performed an experimental determination of the 

12 



increase in the effective frequency of ion collisions in a low-density 
plasma under conditions of ion cyclotron resonance in strong, electric, 
variable fields. It was found that the dependence obtained experimentally 
for the frequency of ion collisions on the high frequency field strength 
cannot be explained on the basis of a theory postulating ion collisions 
with neutrals. The assumption was advanced that such a dependence is 
caused by 'bunched" instability which arises in strong variable fields. 

We can clarify the nature of this absorption by comparing the data 
obtained experimentally, regarding high frequency energy absorption by 
a cold plasma close to ion cyclotron resonance, with the theory advanced 
in (Ref. 5). Let us establish a relationship between the effective fre- 
quency of ion collisions and anomalous plasma diffusion in ion cyclotron 
resonance, which was studied in (Ref. 2). Finally, by making certain 
numerical estimates, we can show that both high frequency energy absorp- 
tion by the plasma, and anomalous plasma diffusion in fields with super- 
critical strength values, may be caused by "bunched' instability produced 
in the field of an ion cyclotron wave (this instability was investigated 
theoretically in [Ref. 8]). 



Description of the Apparatus . 
Measurement Methods 

The apparatus which was employed for the study was described in de- 
tail in (Ref.l). The plasma was produced by pulse discharge with oscil- 
lating electrons in hydrogen, in a 7'10~^— 7 n/m^ pressure range. The 
diameter of the glass discharge tube was 6 cm, and the distance between /18 
the cathodes was 80 cm. The strength of the longitudinal, quasi-constant 
magnetic field could be changed between 4*10'* — 6.4-10^ a/m. 

High frequency energy was introduced into the plasma by means of 
an artificial LC-line which was connected to the self -excited oscillator with 
a frequency of 7. 45 -10^ cps. During excitation at this frequency, 2.5 wave- 
length oscillation was applied to the section of the line x^hich was slipped on 
to the discharge tube. This corresponded to an axial high frequency field 
period of 23 cm. By changing the anode strength of the oscillator, and also 
the connection between the line and the oscillator, it was possible to change 
the amplitude of the high frequency azimuthal line current within 0.5 - 35 
a/ cm. 

The oscillator was switched on for approximately 100 microseconds 
after the pulse of the discharge current had terminated. As is shown in 
(Ref. 1), the maximum high frequency power which could be absorbed by a 
plasma in resonance was 18 kw for a plasma density of 1.7'10'^^ cm"^, and 
an azimuthal line current of 30 a/cm. 

This article presents the measurements of plasma density, electron 
temperature, and the high frequency power absorbed by the plasma. The 

13 



■^ 








* 


0.3 






f 


' 


I 








/ 






0.2 
0,1 




/ 






y 


/ 






/ 









/ 2 3 N-10,cn 

Figure 1 



13 



electron plasma density and its change with time in the (1.7 - 0.25) 10 
range was measured by means of an interferometer at a wavelength of 



cm 



-3 



8.1 mm = 0.81 cm. The experimental point corresponding to a density of 



„-3 



3*10 ^ cm was obtained by extrapolation of the plasma density dependence 
on time toward large densities (Figure 1) . 



The plasma electron temperature was determined according to the de- 
pendence of the luminosity intensity of the line Hg on the high fre- 
quency LC-line current [see (Ref. 2)]. The luminosity intensity of the 
line Hg was measured by means of a UM-2 monochromator and a photoelectron 

multiplier. Due to anomalous diffusion, the plasma decay time, for line 
currents on the order of 5 a/cm and above, was comparable with the electron 
lifetime between two collisions leading to excitation of a neutral atom. /19 
For this reason, for line currents which were greater than, or approxi- 
mately equal to, 5 a/cm, the electron temperature which was computed 
according to the line intensity Hg could be too low. 



The high frequency power absorbed by the plasma was measured with an 
all purpose meter of transmitted power which was described in (Ref. 13). 
The recorders for the current and the strength were connected to the line 
at the point where it was attached to the solenoid. By means of two inter- 
changeable current recorders with different sensitivity, it was possible to 
measure the power absorbed by the plasma, corresponding to a line current of 
0.5-10 a/cm. 

All of the measurements described in this article were performed at 
a hydrogen pressure of 0.133 n/m^. 



14 



Nature of High Frequency Energy Absorption 

Under the conditions of our experiment, the high frequency power 
absorbed per unit of plasma cylinder length is (Ref. 5) 



^ = S ^' (* • ^) (?)'«'. i"J W s5n^& ,z, (3) 



where j q is the amplitude of the azimuthal high frequency current in the 
coil per unit of cylinder length; R — coil radius; a — plasma radius 
(it is assumed to equal the inner radius of the discharge tube in all of 

the computations); Ki — McDonald function; kn = T — — axial wave number 

(it is determined as the axial period X of high frequency current in the 

coil); 0) — oscillator frequency; nn = ; f(X) — the function whose 

specific form depends on the nature of the high frequency energy absorp- 
tion. In particular, for collision absorption we have 

/ {X) = lef f_i (4) 

where 2 



Strictly speaking, relationship (3) is only valid in the case of /2Q 
long-wave fluctuations (or a small plasma filament radius), when k || a«l 

and kj_ a << 1 (k_|_ — radial wave number). Under our conditions, both of 

these quantities are on the order of unity. However, as the computation 
showed, the error produced when equation (3) is applied to our case is 
small . 

Under the conditions of the described experiment, high frequency 
energy absorption by a plasma close to o) = w^ is "collision absorption" 

in the sense that it is described by (3), where f(X) has the form of (4). 
Formulas (3) and (4) may be used to determine the fact that, in the case 
of a hydrogen plasma, the absorbed power is at a maximum in the case of 

^=^ = 2.44.10-'7XW,-. (5) 

Ho 
where 5 = — ; Hq is the magnetic field corresponding to cyclotron 



15 




15 NIO%-' 



resonance of ions at the oscillator frequency; B^ — the magnetic field 
forwhich the high frequency power absorbed by the plasma is at a maximum. 

The dependence described by (5) is shown in Figure 1 (straight line) . 
This figure also presents the points computed on the basis of experimental 
data regarding the shift in the absorption maximum from Hg for different 

plasma densities measured by an interferometer for a line current of 
4.5 a/cm. Over the entire range of measured plasma densities, the devia- 
tion of the points obtained experimentally from the computed dependence 
does not exceed the limits of measurement errors. 



In order to determine what collisions cause the observed high fre- 
quency energy absorption, a study was made of the dependence of the 
effective frequency of ion collisions on the neutral gas pressure and on 
the plasma electron density. It can be seen from (3) and (4) that the 
effective frequency of ion collisions Ypff can be expressed either as 

1 e 

— • — — AH — where AH is the halfwidth of the resonance absorption curve 

2 Mc 



AXUi 



— or as 



.2 



jg, where A is the factor in front of jgfCX) in (3); S^ — 



m 



the power at the absorption maximum. 

It was shown in (Ref . 1) that the power at the absorption maximum 
and the width of the resonance absorption curve depend slightly on the /21 
neutral gas pressure in the 0.133 - 1.33 n/m^ range, due to which fact 
the observed power absorption cannot be caused by ion collisions with 
neutrals . 



Figure 2 presents the dependence obtained experimentally of the 
effective frequency of ion collisions Ypff on the plasma electron density 

for a line current 4.5 a/cm. The crosses designate the points computed 



16 



Y< 


= f 


fio'^ 


3« C 


Fi 










1 


^ 




- 














3 














^ 


- 












2 












^V, 


~v 


e 


Tl- 


• 




/ 


























as 

Ofi 

a^ 

0.2 












. — 




— 


— 


- 






1 


/ 






































1 


















^ 





















T 
60U 


.microsec 










1 


\ 


















500 

400 

^ 300 


1 




V 


-- 


- 


































ZOO 
100 


■r^ 


*-< 



















0//'34S6739 /a,a/c« 

Figure 3 



according to the halfwidth of the absorption curves; the dots designate 
the points computed according to the absolute power at the absorption maxi- 
mum for a given density. The observed high frequency energy absorption is 
caused by ion-electron collisions. However, under the conditions for 
which the dependence shown in Figure 2 was determined, the electron tem- 
perature comprised 1 ev in order of magnitude (see Figure 3) . The frequency 
of ion Coulomb collisions with electrons must be one order of magnitude 
less than Ypff over the entire range of measured plasma densities. There- 
fore, the effective frequency of ion collisions, which was measured in the 
experiment, cannot be caused by Coulomb scattering of ions by electrons 
[the electron temperature, which was determined in (Ref . 1) according to 
Y ff under the assumption of Coulomb interaction of ions with electrons, /22 



amounted to 0.15 ev) . 

In order to clarify the nature of the interaction between ions and 
electrons, and consequently the nature of the observed high frequency 
energy absorption by a plasma in ion cyclotron resonance, the dependence 
of the effective frequency of ion collisions on the line current density 
JO — i.e., the dependence on the strength of the high frequency field in 
the plasma — was determined. The quantity Ygff ""^^ computed according to 



17 



the power at the absorption maximiun for a density of 5.1 •10-'^^ cm ^. 
Figure 3 shows the dependences Ygff (Jo)» '^^(■20^ a^^i t(jo) — where t is 

the plasma decay time — on density 7. 6*10^^ up to 2.5'10-'-^ cm~^ for a 
magnetic field strength of 4.2»10^ a/m, where the absorption reaches a 
maximum for a density of 5.1*10^^ cm"^. It can be seen that Yoff rapidly 

decreases with an increase in joCjo*^ 1 a/cm) . The frequency of Coulomb 
collisions between ions and electrons must follow this pattern, since the 
high frequency power introduced into the plasma — and consequently the 
electron temperature — increases with an increase in j q . The electron 
temperature, determined according to Yeff ^^ ^^^ case of Jq< 1 a/cm under 

the assumption of Coulomb interaction between ions and electrons, coin- 
cides in order of magnitude with the temperature computed according to 

the line intensity H (see Figure 3) . 

p 

There is a sharp bend in the curve y ff (Jo) ^^ the point Jo ^ 1 a/cm, 

and with a further increase in jg, y^ff increases slowly, remaining at a 

level on the order of 2*10^ sec~^ and considerably exceeding the frequency 
of Coulomb collisions between ions and electrons for given electron tempera- 
tures. As can be seen from Figure 3, at this point a sharp decrease in the 
plasma decay time occurs. It was shown in (Ref . 2) that a decrease in the 
decay time is caused by increased plasma diffusion across the force lines 
of the magnetic field. In its turn, the diffusion is caused by an insta- 
bility produced in the field of an ion cyclotron wave having a large ampli- 
tude. 

It may thus be assumed that under the conditions of our experiment 
the anomalously large frequency of ion collisions in high frequency fields 
with supercritical strengths is caused by the more intense (as compared 
with Coulomb interaction) interaction of ions with electrons. The reason 
for this (and for anomalous diffusion) is "bunched" instability. The fact 
that absorption is "collision absorption" — i.e., it fomnally satisfies 
equation (3) — where f (x) in the form of (4) means in physical terms that 
the damping force (which is caused by an instability) of the ion motion /23 
directed toward the electrons is proportional to the relative velocity of 
ion and electron liquids in the field of an ion cyclotron wave. 

Comparison With the Theory of ^on 
Cyclotron Wave Stabi l i ty^ 

We shall show that the values of the effective frequency of ion colli- 
sions in fields with supercritical strengths, computed either according to 
the halfwidth of the resonance absorption curve or according to the absolute 
power at the absorption maximum, as well as the plasma diffusion coefficient 
determined by the decay time, coincide in order of magnitude with the corres- 
ponding values computed for our case according to the theory of ion cyclotron 
wave stability (Ref. 8). 

18 



As follows from (Ref. 8), under the experimental conditions (Ng~ 
~5*10^2 cm~^; H~ 4*10^ a/m; Tg ~ 1 ev) for longitudinal (electrostatic) 

high frequency fluctuations excited by an ion bundle and propagated almost 
across a constant magnetic field (cos 9 :$, 2.) , we obtain 

Reu) — Imo) — (u>e"Jj)'^'> (6) 

where o) is the cyclotron electron frequency. 

The effective frequency of ion collisions may be regarded as the in- 
verse of the time during which an ion bundle with an initial energy per 

unit of volume N^ — r— (u — the velocity of the relative motion of the ion 

and electron components in the field of an ion cyclotron wave almost equals 
the velocity of an ion liquid) excites the fluctuations (6) and transmits 

all of the energy of the ordered motion to the electron gas. On the other 

2 
hand, the electrons obtain the energy Ni from the ions during the ex- 

bundle is thus 



citation time of the fluctuations (u oj.) '^. The braking time of an ion 



-:=— (u),o),) 2. (7) 



m 



Substituting numerical values in (7), we obtain 10 ° sec for x, which coin- 
cides in order of magnitude with Ylf f ~ 0.5'10~^ sec which was determined 
experimentally. 

Let us determine the diffusion caused by instability, employing the /24 
theory of nonuniform plasma stability (Ref. 14). The diffusion coefficient 
may be written as follows 

D^V^(, (8) 

where V is the plasma pulsation velocity; t = — — the characteristic time 

of correlation disappearance. The increasing increment c£ fluctuations (6) 
must be used as v. The pulsation amplitude may be determined by the condi- 
tion of balance between two processes — one of which leads to an increase 
in the pulsation amplitude due to the development of instability, and the 
other leads to contraction of this amplitude due to nonlinear processes 
leading to oscillation damping. As a result, we obtain 

y = vx^, (9) 

where Xj_ is the oscillation wavelength in the radial direction. Then the 



19 



diffusion coefficient is 

£>~vXi, (10) 

or, since ku ss v, where k is the radial wave number, we have 

D~(2u)«^. (11) 

Under the experimental conditions, in the case of jo = 8 a/cm, u ~ 
~ 10 cm/sec and D ~ 2'10 cm^/sec, which coincides in order of magnitude 
with D ~ 5 '10^ cm^/sec, determined for the same conditions according to 
the plasma decay time in (Ref. 2). 

Conclusions 

1. The absorption of high frequency energy, which was observed in 
our experiments, close to the ion cyclotron resonance both for small and 
large amplitudes of the high frequency field can be formally described by 
relationships (3) and (4), which were derived for the case of "collision" 
absorption. 

2. Absorption is caused by the interaction between plasma ions and 
electrons. In fields whose strength is less than the critical strength, 
it is caused by pair Coulomb collisions. In fields whose strength is 
greater than the critical strength, the effective frequency measured 
experimentally of ion collisions is considerably greater than the fre- 
quency of pair Coulomb collisions. In this case, the anomalously large /25 
absorption is apparently caused by braking of the ordered motion of ions 

in the field of the ion cyclotron wave by high frequency longitudinal 
oscillations excited by an ion bundle. 

3. The value obtained experimentally for the effective frequency 

of ion collisions, and also the diffusion coefficient determined according 
to the plasma decay time, coincide in order of magnitude with the corres- 
ponding values calculated according to the theory of ion cyclotron wave 
stability. 

4. On the basis of the results obtained, it is natural to pose the 
question of "turbulent" plasma heating (Ref. 15) in ion cyclotron resonance 
— i.e., brief excitation in the plasma of an ion cyclotron wave having a 
large amplitude, with subsequent thermalization of the ordered motion of 
plasma particles in the field of this wave by the high frequency longi- 
tudinal oscillations excited by an ion bundle passing through the electron 
gas. The heating period must be quite small, so that anomalous diffusion 
caused by plasma instability does not produce significant losses in plasma 
particles. 



20 



REFERENCES 

1. Vasil'yev, M. P., Grigor'y^va, L. I,, Dolgopolov, V. V., Smerdov, B. I., 

Stepanov, K. N. , Chechkln, V. V. Zhurnal Tekhnicheskoy Flzlki, 34, 
6, 1964. 

2. Chechkin, V. V., Vasil'yev, M. P., Grlgor'yeva, L. I., Smerdov, B. I. 

Yadernyy Sintez, 4, 2, 1964. 

3. Sagdeyev, R. Z. , Shafranov, V. D. In the Book: Plasma Physics and 

the Problem of Controlled Thermonuclear Reactions (Fizika plazmy 1 
problema upravlyayemykh termoyadernykh reaktsiy) . Izdatel'stvo AN 
SSSR, Moscow, 430, 1958. 

4. Stix, T. H. Phys. Fluids, 1, 308, 1958. 

5. Vasil'yev, M. P., Grlgor'yeva, L. I., Dolgopolov, V. V., Smerdov, 

B. I., Stepanov, K. N., Chechkin, V. V. Zhurnal Tekhnicheskoy 
Flziki, 6, 34, 1964; In the Book: Plasma Physics and Problems of 
Controlled Thermonuclear Synthesis, 3 (Fizika plazmy i problemy 
upravlyayemogo termoyadernogo slnteza, 3). Izdatel'stvo AN USSR, 
Kiev, 96, 1963. 

6. Kadomtsev, B. B. In the Book: Plasma Physics and the Problem of 

Controlled Thermonuclear Reactions (Fizika plazmy i problemy uprav- 
lyayemykh termoyadernykh reaktsiy), 4. Izdatel'stvo AN SSSR, Moscow, 
364, 1958. 

7. Kurilko, V. I., Miroshnichenko, V. I. In the Book: Plasma Physics and 

Problems of Controlled Thermonuclear Synthesis, 3 (Fizika plazmy i 
problemy upravlyayemogo termoyadernogo sinteza, 3). Izdatel'stvo 
AN USSR, Kiev, 161, 1963. 

8. Stepanov, K. N. Zhurnal Tekhnicheskoy Fiziki, 34, 12, 1964. 

9. Hooke, W. M. Nuclear Fusion, Supp., 3, 1083, 1962. 

10. Nazarov, N. I., Yermakov, A. I., Tolok, V. T. Present Collection, 1. 

11. Nazarov, N. I., et al. Plasma Physics and Problems of Controlled /26 

Thermonuclear Synthesis, 3 (Fizika plazmy i problemy upravlyayemogo 
termoyadernogo sinteza, 3). Izdatel'stvo AN USSR, Kiev, 164, 1963. 

12. Dubovoy, L. V., Shvets, 0. M. , Ponomarenko, A. I. In the Book: Prob- 

lems of Magnetic Hydrodynamics (Voprosy magnitnoy gidrodinamiki) 2. 
Izdatel'stvo An Latviyskaya SSR, Riga, 355, 1962. 

13. Yakovlev, K. A., Pankrushina, D. K. , Basin, Yu. G. Pribory 1 Tekhnika 

Eksperimenta, 4, 89, 1961. 



21 



14. Galeyev, A. A., Moiseyev, S. S., Sagdeyev, R. Z. Atomnaya Energiya, 

15, 6, 451, 1963. 

15. Zavoyskiy, Ye. K. Atomnaya Energiya, 14, 1, 57, 1963. 



INVESTIGATION OF CONDITIONS PRODUCING A DENSE PLASMA IN A 
METALLIC CHAMBER AND ITS HIGH FREQUENCY HEATING 

0. M. Shvets, S. S. Ovchinnikov, V. F. Tarasenko, 
L. V. Brzhechko, 0. S. Pavlichenko, V. T. Tolok 

It was shown in (Ref . 1) that powerful, high frequency oscillators may 
be employed to produce a dense plasma in a metallic chamber. The present 
experiments represent a continuation of this study by investigating the 
conditions producing a dense plasma in a metallic chamber and by determining 
the main plasma parameters. 

The utilization of a metallic chamber as the operational vacuum body 
and the method of feeding the system from the hf oscillator have definite 
advantages and unusual features. 

1. The system has a low input impedance, which does not require high 
voltages when large, high frequency powers are introduced, and avoids 
several technical difficulties which accompany, as an example, the spatially 
periodic circuit proposed by Stix (Ref. 2). 

2. A good connection between the feed electrodes and the plasma is 
provided. A dense plasma with a cylindrical form is produced between the 
central electrodes, and has no direct contact with the wall of the discharge 
chamber. 



Investigation of the Conditions Producing a 
Dense Plasma in a Metallic Chamber 

The "Vikhr"' device (Figure 1), on which the study was performed, con- 
sists of a copper chamber 1 with a wall thickness of 2.5 mm, an inner diameter 
of 125 mm, and a length of 2000 mm. It is placed in a magnetic field which 111 
can be regulated continuously between -2*10^ a/m. The configuration of the 
magnetic field may be varied, depending upon the method chosen to handle 
the plasma produced. Since we were interested in plasma heating by high 

22 




Figure 1 



frequency fields , the magnetic field configuration was chosen in accordance 
with the conditions which were requisite for generation and absorption of 
ion cyclotron waves. The region of the "magnetic beach" was located in 
the center of the selenoid 11 producing the magnetic field. The residual 
gas pressure in the system did not exceed 1.33-10"'* n/m^. The rods 3 are 
introduced axially into the vacuum chamber through the porcelain insula- 
tors 2. Aluminum electrodes 4 with a diameter of 50 mm and a length of 
70 mm are located at the ends of these rods. The distance between the 
electrodes is 1000 mm. The feed rods 3 through the coaxial cables 5 and 
the capacitance C are connected with the coupling coil 6 of the hf oscil- 
lator 7. The oscillator power is on the order of 100 kw; the operating 
frequency is 1.82' 10^ cps. The ends of the chamber are closed with glass 
discs 8. The operational gas from the flask 9 is admitted into the chamber 
by the valve 10 through an opening in the glass disc 8. 

With the system employed for switching on the hf oscillator, the feed 
electrodes acquire a negative potential, due to the rectifying properties 
of the plasma; this negative potential can reach several kilovolts with 
respect to the chamber. This potential creates the condition for oscilla- 
tion of electrons along the magnetic field force lines between electrodes, 
similarly to Penning discharge. The oscillating electrons effectively /28 
ionize the operational gas and provide for a high degree of ionization up 
to a chamber pressure of 1.33*10-^ n/m^. A core of dense plasma is formed 
between the electrodes along the chamber axis; the diameter of this plasma 
is determined by the electrode diameter. Due to the presence of a constant 
potential, the peripheral plasma begins to rotate according to the law of 
plasma behavior in crossed electric fields and magnetic fields. In a few 
microseconds the applied high frequency voltage produces a discharge in /30 
the operational area, which is accompanied by a voltage decrease at the 
electrodes due to an increase in the oscillator loading (Figure 2 , a) . 
The central electrode acquires a negative potential, which is retained for 
the entire period of time. that the dense plasma exists (Figure 2, b) . 



23, 








w ■ 



9n i 



msec 



Figure 2 



Lines of admixtures from the chamber wall materials and from the 
electrodes — Cull, AIII — and a weak line CII (Figure 3) were observed 
in the discharge spectrum obtained by means of a ISP-51 spectrograph with 
a UF-85 camera (focal distance 1300 mm) . 

When the outline of the line H„ was measured, it was found that under 

p 

our experimental conditions there was an apparant Stark widening by the 
micropoles of the plasma. This made it possible to determine the charged 
particle density, by comparing the contour observed experimentally with 
the theoretical contour computed on the basis of the theory advanced by 
Kolb, Grim, and Shen (Ref. 3). For purposes of comparison. Figure 4 shows 
the observed contour I and the theoretical Stark contour II , computed for 
n = 2.0*10^'* cm~^. The figure also plots the Gaussian contour III with 
the halfwldth equalling the experimental value of the halfwidth 0.7 A. 
Close to the maximum, the line broadening was caused by the Doppler 
mechanism, and the slopes of the line are due to broadening by the Stark 



24 




/29 



He I 5875 



Cul 5216 
Cui 5153 
Cul 5111 
Cui 5101 
Hei 50^7 
Hei 5015 


Hei mi 


H^ mi 


Cll mtf 
Hei ^713 

Hen me 



Hi ttd^O 



Figure 3 



mechanism. 



Measurements of the plasma density made it possible to determine its 
comparatively weak dependence in discharge on the magnetic field strength, 
which confirmed the assumption regarding the nonresonance mechanism by 
which a plasma is produced in discharge. At a pressure of 0.4 n/m^ , the max- 

cm^ 



imum plasma density was 2*10^ 



For an approximate determination of 



25 




Figure 4 



the plasma density distribution over the radius of the system, different 
sections of the plasma were focused on the spectrograph slit. The most 

>1^ 1 



dense plasma (density on the order of 10^ 



cm 



3" ) produces a filament with a 



diameter of 20 mm; at a distance of about 30 mm from the discharge axis, 
the plasma density decreased by more than one order of magnitude. 

The electron temperature was measured with respect to the intensities /32 
of singlet and triplet lines of helium; it was (4 - 5) 10 °K. 

The dependence of the plasma density change on time was determined 
by probing the plasma with an ultrahigh frequency signal at a wavelength 
of 3 cm (Figure 5, c) and 0.8 cm (Figure 5, d) through the glass openings in 
the chamber. The measurements showed that in an optimum regime there is 

a plasma with a density of > 10 — j in the apparatus for 3.6 milliseconds. 



cm- 



and with a density of > 10 



12 



cm^ 



for 17 milliseconds. The form of the hf 



pulse is shown in Figure 5, a. The maximum plasma density was determined 
according to measurements of the relative change in the intensity of the 

spectral line H. with time. The intensity of the spectral line H„ (the 

p p 

photomultiplier signal. Figure 5, b) , if the main excitation mechanism is 
electron collision, is 



h 



•^4 2 + ■^41 



where v^2 is the frequency corresponding to the line H ; A., — probability 
of spontaneous transition from the i level to the k level; 

< vai (u) } = f voi {v) fe {v) dv; 





26 




/31 







w 



. — -> ,-,. 



i^U t, msec 



Figure 5 



a. (v) — excitation cross section (by electron collision) of the n = 4 
level. fgCv) is the electron distribution function with respect to 

velocity; [v^^Cv)] in the case of Maxwell distribution of electron veloci- 
ties is the function of electron temperature which — as measurements of 
the dependence T (t) have shown — changes very little during the dura- 
tion of the hf pulse. Thus, the intensity of the line Hg must change 
proportionally to the product n n In our case, the neutral gas from 

the cold section constantly enters the discharge column, since 
Rpl 

^n 



10~^ sec (Rpl — radius of the plasma coliimn; v — thermal velocity 



27 



of the neutrals). Therefore, the neutral density n^ « const and, conse- 
quently, the intensity of the line H is proportional to the electron 

p 

density n . At the moment when n = 10-^^ — g-, the intensity of the line 
H decreases by a factor of 200 as compared with its maximum intensity — 

i.e., the maximum plasma density in our case is 2*10-^'*^ — j, which coin- 
cides with the contour of the line H^ determined according to Stark 

p 

broadening. The plasma density determination based on the intensity of 

the line H , at the moment when a signal with a wavelength of 0.8 cm /33 

begins to pass through, coincides with the result of microwave measure- 
ments . 

Thus , we have studied the conditions producing a dense plasma in a 

metallic chamber. A plasma with a density on the order of 10^** — r 

■' cm^ 

and an electron temperature (4 - 5) 10^ °k was obtained in the experiments. 
The weak dependence of plasma density on the magnetic field strength points 
to a nonresonance mechanism by which the plasma is produced. The aim of 
our subsequent experiments was to investigate the heating of the plasma 
obtained by the generation of ion cyclotron waves. 



High Frequency Heating of a Dense Plasma 
in a Metallic Chamber 

The next stage of our investigation was to study the possible heating 
of ions by generating ion cyclotron waves in a dense plasma produced by a 
hf oscillator in a metallic discharge chamber. The "Vikhr"' device was 
modernized so that, when the ion cyclotron wave was generated at the ends 
of the coaxial close to the magnetic mirror, it was propagated into the 
center of the discharge chamber, where the magnetic field strength was de- 
creased to a value equalling the cyclotron value for protons, forming the 
region of the magnetic beach. When the wave was propagated along the system 
axis and approached the region of the magnetic beach, its velocity decreased, 
and it was damped, transmitting its energy to the plasma ions. In order 
to propagate the wave in the region of the magnetic beach, it is necessary 
that the magnetic field strength per wavelength change by several percents 
— i.e., a smooth change in the magnetic field strength over the length of 
the system is requisite. If the opposite is true, the wave will be reflected. 

The resid-ual gas pressure in the system did not exceed I.S'IO"** n/m^. 
The experiment was performed with a mixture of two gases — hydrogen and 
helium or hydrogen and argon. The operational pressure was established 
by a stationary regime of the valve operation in the 0.1 - 0.8 n/m^ 
range. At a pressure of 0.4 n/m^, the hf oscillator, operating at a 



28 



frequency of 1.82 '10^ cps with a power of 150 kw, produces a plasma with 

a density on the order of 10-^^ j. With a magnetic field strength of 

H = 1.25 H. (H. — the cyclotron value of the magnetic field strength 



x.c i.c 



A. 



for protons) waves are generated in the plasma with X = = 6.6 cm /34 



(A — wave length in a vacuum: n= f ' i o> \2 — refractive index 
vac ^ ' , / - ( 1 



i; n= f ' / 0. \2 



TJ 

of the medium; v. = -—^^^ — Alfven velocity; H — magnetic field; u) — 

operational frequency of the oscillator; w. — cyclotron frequency of 
proton rotation; p — mass density. 

This system has several advantages. The pulse which is initially 
transmitted to the ions is perpendicular to the vector of the outer mag- 
netic field strength, which facilitates the retention of ions in the cork- 
screw configuration of the magnetic field. Since the system has a low inped- 
ance, energy is readily introduced into the discharge chamber and is trans- 
mitted directly to the plasma ions. An increase in the density and diameter 
of the plasma does not make the conditions worse for wave generation. At 
the same time, the spatially periodic circuit for Introducing hf power 
into the plasma, which was advanced by Stix (Ref. 2), loses any physical 
meaning with an increase in the plasma density and diameter. When a mix- 
ture of two gases is heated, or when it is necessary to heat the plasma 
electron component simultaneously with the ion component, this system makes 
it possible to introduce the power of two oscillators operating at different 
frequencies. 

As has been pointed out , an Increase in the plasma density does not 
impede wave generation, since spatial periodicity is not given externally, 
but is established as a function of the plasma refractive index, and may 
be small (several centimeters). However, the possibility of periodicity 
is not excluded, if rings are placed endwise on the coaxial at different 
distances from each other; these rings will introduce a perturbation, 
creating a specific periodicity along the system axis. One unusual feature 
of the discharge is the fact that its nucleus, consisting of a plasma which 
is almost entirely ionized, is surrounded by a plasma having a low density 
and a neutral "housing" . During generation and absorption of ion cyclotron 
waves, the dense plasma nucleus is heated, and the cold plasma surrounding 
it with a low density contributes, in all probability, to the suppression 
of channel instability. 

In the experiments described, the ion temperature of the plasma and 
the charged particle density, which were averaged over time, were deter- 
mined by means of optical methods. The change with time in the intensity 
of the spectral lines for hydrogen and admixtures was also studied. A 
ISP-51 spectrograph was placed in such a way that the central portion of /35 



29 



^^^^''d- 



Figure 6 



the cylindrical plasma column, lying in the region of the magnetic beach, 
was focused on its input slit. Oscillograms showing the intensity of 
the spectral lines for hydrogen and helium or argon, added to the chamber 
in small proportions to the operational gas, showed that regular oscilla- 
tions are produced in the intensity of the spectral lines with a frequency 
on the order of 20 kc, close to the magnetic field strength corresponding 
to the cyclotron value for protons (Figure 6) . These oscillations appear 
during wave generation in the plasma, and are caused by the eccentric ro- 
tation of the dense plasma filament as a whole with respect to the chamber 
axis, in accordance with the drift law in crossed, radial electric fields 
and axial magnetic fields. The direction and frequency of the filament rota- 
tion was determined by two photoelectron multipliers oriented towards the 
end of the chamber and placed at a radius of 3 cm from the system axis. 
One of them was shifted along the azimuth. The oscillation phase of the 
light intensity was thus changed. 

One interesting feature was discovered when the ion temperature was 
measured according to the Doppler broadening of the hydrogen and additional 
gas lines . The width of the hydrogen line depended comparatively little 
on the magnetic field strength, and the additional gas lines were broadened 
considerably when the magnetic field strength was close to the cyclotron 
value for protons. The ion temperature of this gas, determined for the /36 
optimxun operational regime of the apparatus, amounted to 250 ev (2.5*10^ °K) . 
Measurements of the halfwidth of the line H. showed that the hydrogen atom 
temperature was below the temperature of the additional gas, while there 
was considerable Stark broadening of the line contour, corresponding to a 

ai+ L 



plasma density of 2*10 



cm- 



There was also Doppler broadening of the 



admixture lines (copper, alvuninimi, oxygen, carbon, nitrogen). When the 
temperatures of different admixtures and additional gases were measured. 



30 



l-'t 




■* 8 12 16 H-10":alt^i 
Figure 7 



no Ion temperature dependence on their mass was detected. The heating 
of the additional gas ions was terminated when only one additional gas 
was admitted into the discharge chamber up to the previous operational 
pressure. One of the characteristic dependences of helium temperature 
(when it was added to the chamber in a small amount) on magnetic field 
strength, measured over the half width of a helium line with a wavelength 
of 4921.93 A, is shown in Figure 7. The helium atoms acquire a maximum 
temperature at a magnetic field strength which is close to the cyclotron 
value for protons. A certain temperature increase is observed as a mag- 
netic field strength is approached which equals the double cyclotron value 
for hydrogen ions. 

If it is assumed that the halfwidth of the H_ line is determined by 

p 

Doppler broadening, the temperature of neutral hydrogen is considerably 

lower than the plasma ion temperature. However, under our conditions 

(in the case of T = [4-5] 10^ °K) , the lifetime of neutral hydrogen in 

the plasma, with respect to the ionization process, was small as compared 

with the time of Coulomb collisions. Therefore, neutral hydrogen cannot 

acquire energy equalling the ion energy. Since the frequency of hydrogen 

atom collisions with electrons is greater under these conditions than the 

frequency of collisions with ions , the Stark mechanism is the predominant 

mechanism leading to the broadening of the line of the residual neutral 

hydrogen. 

In an optimimi operational regime in a mixture of two gases — hydrogen 
and argon, the dependence of the argon ion temperature on the magnetic 
field strength is resonant in nature with a maximirai close to the magnetic 
field strength corresponding to the cyclotron value for protons . The 
maximum temperature of argon ions in the experiments was 2. 5 •10° °K, and 
for electrons — 5 '10^ °K. In order to determine the distribution of the 
plasma ion temperature over the radius of the system, different sections 
of the plasma were focused on the spectrograph slit. The hottest plasma was 
located in the center of the filament. The temperature rapidly decreased /37 
over the radius (approximately 5 times greater along the axis than on a 
radius of about 4 cm) . Figure 8 Illustrates the temperature dependence of 



31 




' '.5 2 U.kv 
Figure 8 



plasma on the applied high frequency voltage. The increase in the mag- 
netic field strength and the high frequency voltage — i.e., the hf power 
introduced — made it possible to obtain a higher plasma temperature. 

The mechanism by which the proton energy is transmitted to the addi- 
tional gas has still not been definitely clarified. Since the gasokinetic 
pressure in these experiments may exceed the magnetic pressure, the possi- 
bility is not excluded that centrifugal instability may be produced, which 
can lead to transmission of proton energy to the additional gas and can 
lead to its heating. 

In certain operational regimes of the apparatus, we observed genera- 
tion of a rapid magnetosound wave at a magnetic field strength which was 
less than the cyclotron value for protons. However, conditions were not 
favorable for studying it at the existing oscillator frequency (1.82-10° 
cps) . 

Thus, in all probability, these experiments illustrate the feasibility 
of high frequency heating of a dense plasma consisting of two types of 
ions by resonance generation of ion cyclotron waves for one type of ions. 
The mechanism by which energy is transmitted from one type of ions (pro- 
tons) to other ions (helium, argon, admixture) requires further study. 

REFERENCES /38 

1. Shvets, 0. M. , Tarasenko, V. F., Ovchinnikov, S. S. , Tolok, V. T. 

In the Book: Plasma Physics and Problems of Controlled Thermonuclear 
Synthesis (Fizika plazmy i problemy upravlyayemogo termoyademogo 
sinteza), 3. Izdatel'stvo AN USSR, Kiev, 117, 1963. 

2. Stix , T. and Palladino, R. In the Book: Transactions of the Second 

International Conference on the Peaceful Utilization of Atomic 
Energy (Trudy Vtoroy mezhdunarodnoy konferentsii po mirnomu ispol'- 
zovaniyu atomnoy energii) . Geneva, 1958. Selected Reports of 



32 



Foreign Scientists. Physics of a Hot Plasma and Thermonuclear 
Reactions (Izbrannyye doklady inostrannykh uchenykh. Fizika gorya- 
chey plazmy i termoyademyye reaktsil.) Atomizdat, Moscow, 242, 
1959. 

3. Grim, H. R. , Kolb, A. C, Shen, K. J. Stark Broadening of Hydrogen 
Lines in Plasma, Phys. Rev., 116, 4, 1959. 



HIGH FREQUENCY PLASMA HEATING 

K. N. Stepanov 

As is well known, a rapid decrease in the frequency of Coulomb colli- 
sions of plasma particles leads to a decrease in the rate of ohmic plasma 
heating both by constant fields and by high frequency electric fields when 
there is a temperature increase. Therefore, great hopes have been expressed 
(Ref. 1, 2) for achieving plasma thennonuclear temperatures by utilizing 
different effects of collislonless energy absorption of high frequency 
fields by a plasma. In principle, this amounts to Cherenkov, or to cyclo- 
tron, absorption and radiation of waves by electrons and ions. 

However, collislonless energy absorption by a plasma leads to great 
distortion of the velocity distribution function of electrons and ions and 
to attenuation, or even complete discontinuance, of absorption. It also 
leads to every type of plasma instability. If collislonless plasma heating 
is effectuated by weak electric fields, so that Coulomb collisions provide 
a Maxwell distribution function, the heating rate is the same as for ohmic 
heating. The time required to heat the plasma up to thermonuclear tempera- 
tures with weak fields is less than the time required for containing the 
plasma in a thermonuclear reactor, utilizing the reaction D + D, with a 
positive energy output. Therefore, under the condition of producing a 
stable, plasma configuration, it is primarily possible to achieve thermo- 
nuclear plasma temperatures with slow heating by weak, constant fields or 
by high frequency fields, which in themselves do not lead to strong plasma /39 
instability. 

On the other hand, for rapid plasma heating it is very tempting to 
employ strong electric fields, under whose influence electrons or ions 
in the plasma acquire a large directed velocity (Ref. 3-6), as well as 
strong electron bundles (Ref. 7) or the collision of plasma clusters 
(Ref. 3) ("turbulent" methods of plasma heating). Due to the development 

33 



of bunched Instability in the systems, there is a rapid increase in the 
energy of high frequency oscillations leading to braking of the acceler- 
ated particles, to an energy increase with respect to motion ("tempera- 
ture"), and to an energy exchange between different plasma components. 

This article presents a brief survey of the collisionless, high fre- 
quency methods of plasma heating, and it also compares them with the 
ohmic heating method. 



Ohmic Plasma Heating 

When investigating the processes of plasma heating, we shall disre- 
gard energy losses due to cyclotron radiation of electrons, which is ab- 
sorbed by a plasma if its dimensions are fairly large, as well as losses 
due to plasma thermal conductivity on the chamber walls — which is also 
a surface effect. In addition, we shall assume that the residual gas 
pressure is small, so that we can disregard energy losses due to over- 
loading and radiation of admixed atoms excited by electron collision. 
Under these conditions, the energy losses by the plasma are determined 
only by braking radiation of the electrons . 

When the ohmic method is employed to heat a plasma by a "constant" 
electric field with the strength E, the electron energy increase is deter- 
mined by the following equation 

§=^Q+-Q-, (1) 

3 
where w = y T is the mean electron energy; T — electron temperatures; 

n,Q+=aE^ (2) 

— Joule heat liberated per unit volume per unit of time; ng — electron 
density; a = _^ e — plasma conductivity; ^ = ^ '"'''' — time of /40 

electron mean free path; A — Coulomb logarithm. The intensity of elec- 
tron braking radiation (Ref . 8) is 

32 yW.e« 

Q— = - - a) 

If we take into account the transmission of energy to ions , then we 

must replace Q, by -r Q, . in the case of T^ « T^. This effect, as well as 

other effects which do not change in order of magnitude, is not taken 
into account from this point on. 

34 



Under the Influence of the electric field E, the electrons acquire 
a directed velocity 



u = 



enoVg 



If the field strength E exceeds the critical value (E ~ , where 

a 

y«=l/ ^ is the electron thermal velocity), then all of the electrons 

are carried along by the electric field in a continuous acceleration regime. 
The motion of the electron gas with respect to the ions leads to the phe- 
nomenon of bunched instability which is related to the buildup of longi- 
tundinal high frequency oscillations in the plasma (Ref. 9). The inverse 
influence of plasma oscillations on the electron motion leads to electron 
braking — i.e., to anomalous plasma resistance, and also to increased 
radiation of radio waves, which considerably increases the plasma thermal 
radiation (Ref. 10, 11). 

Plasma bunched instability can arise in the case of E << E^^, i.e., 
in the case of u << Vg. For example, in a very non-isothermic plasma, 
in the case of T^ >> Tj^ (T- — ion temperature) the electrons build up 
sound oscillations (Ref. 12), if the velocity u exceeds the speed of sound 

Vs = l/ ^ . In the case of T < T^, the electrons build up longitudinal 
ion cyclotron oscillations, if u > 10 (v^ = l/ — ■ — thermal ion velo- 
city) (Ref. 13). However, it may be expected that — since only a small 
group of resonance electrons participates in the oscillation buildup — 
the formation of a "plateau" In the electron distribution function will /41 
lead to a decrease in the Increasing increment ( Ref. 14 - 18), and nonlinear 
effects will lead to stabilization of these oscillations (whose amplitude 
will be small) (Ref. 16 - 18). 

Expression (2) may be employed only in the case of E << E^^.. Assuming 
that E = aE^j-, where a << 1, we obtain 

Q+ ^' (4) 

Let us substitute expressions (3) and (4) in (1) . We then have 



dt 



V T. 



5-lG-'5Kr,jnokev/sec. (5) 



35 



where Tg — in kev, ng — in cm ^. The heating process terminates, 

-J— = in the case of 

T —S-lO^a^kev 
max . ,,^ 

It thus follows that even in the case of heating with weak fields (a = 
= 10" -"^ - 10~^) it is possible to achieve thermonuclear temperatures 
(T -x. 50 kev) . 

If Q_ << Q, , then — disregarding Q as compared with e Q in (1) 
and taking into account (4) — we obtain 



n = 7-.(i + "^)"'-. (7) 



where Tq and tq — are the initial values of T^ and x . The influence of 
braking radiation on the heating process is significant in the case of 
Tg 0, Vax' i*^*» ^^ the case of 

/5- WV/i 
^-^"^[-f^j ■ (8) 

For example, in the case of Tg '\. 100 ev, ng 'v 10^^ cm~^, Tg 'v^ 2"10 ^ 
sec and a '^ (1/30), the temperature T = 50 kev is achieved during the 

time t '^ 0.2 sec 

'-AtJ ' (9) 



and the electric field strength changes between 0.3 - 3 '10"^ v/cm. The 
energy exchange between ions and electrons takes place during the time 
T^g ~ 0.5 sec 

mi 
^'•«~^^m, • (10) 

so that the separation between the electron temperature and the ion tem- /42 
perature is small. 

The heating time (9) is not large from the point of view of producing 
thermonuclear reactors with a positive energy balance. The time the plasma 
is contained in such a reactor, when employing the reaction D + D, must be 
greater than (Ref. 19) 

t*—-^ sec. (11) 



36 



For the example under consideration (ng '^ 10-^^ cm~^) t* "^ 10 sec, 
which exceeds the heating time of the Ion component by a factor of 20. 

Cherenkov Ion Heating 

In the case of high frequency plasma heating employing the method 
advanced in (Ref. 20), oscillations of an axial magnetic field H^ which 
is produced by azimuthal electric currents, lead to a variable azimuthal 
electric field E, in a plasma cylinder located in a strong longitudinal 
field Hq. Radial, drift oscillations of the plasma arise, due to particle 
drift in crossed fields E. and Hq, which leads to density oscillations — 

i.e., to the appearance of sound (more precisely, magnetosound) waves. 

(It is assumed that the wave frequency w is considerably less than the ion 

gyrofrequency ui^ = 2.^ and the wave length is considerably less than the 

m-j^c 

V-f 

Larmor radius of ions p,- = with thermal velocity). 

wi 

Ions having the velocity v,, along Hg , which is close to the phase 

velocity of a wave Va H , vigorously interact with the field E. . If 

•^ II 
the wave phase velocity is on the order of the ion thermal velocity, the 
number of resonance ions is large, and there is strong wave absorption. 
(As is known, in the absence of a magnetic field sound oscillations in a 
plasma, in the case of Tg < Tj_, cannot be propagated in general, due to 

strong Cherenkov absorption by ions (Ref. 21); in the case of Hq ^^ and 
Tg < Tj^, magnetosound oscillations are also damped during one period, if 
V^ ^Vi [Ref. 22]). 

Let us investigate Cherenkov absorption by plasma ions of the energy 
of an electromagnetic field produced by azimuthal electric currents, /43 
which take the form of a moving wave and flow into the coll placed on a 
plasma cylinder having the radius a: 

/^=/„cos(;%,z — W)8(/- — a). (12) 

If k|| a< 1, then the current (12) produces a variable longitudinal 
magnetic field with the strength 

H^=Hcos{k,z — wt), (12') 

where H = — il. The strength of the aximuthal electric field is 



ff 

Ef = — ^— kirs'm{k,z — uif). 



knc 



2n, 



where ni = is the longitudinal refractive index. 



37 



In the absence of damping, the value of the energy flux, averaged 
over time, in the plasma per unit length 



•^0 "^ "■ 4^ Ey/ZzSica 



equals zero. It is apparent that, in the presence of damping, the energy 
flux in the plasma equals, in order of magnitude, 

4«," in^y (13) 

where x is the correction to the "transverse" refractive index nj_, caused 
by the Cherenkov oscillation absorption by plasma ions. 

The dispersion equation for a magnetosound wave has the form (Ref . 22) 

^/, , .-2 "f 

"a 



<- + ^\ =^M+'«-P^' (14) 



where the coefficient n ~ 1, if Vx ~ v^. The component ~ i takes into 
account Cherenkov wave damping in an ion gas . 

Let us investigate a plasma with a small gasokinetic pressure 
/SitnoT,-,, \ ff 

— 772~^C1 • In this case, the Alfven velocity Va=-t== is consider- 

\ "o J y4jinom.. 

ably greater than the ion thermal velocity v^^ and the speed of sound Vg. 

Since nn ~ — , in the zero approximation it follows from equation (14) that /44 

nj_ = in||. In the following approximation we obtain: "j. =ini (l + y '"^j ^^)' 



I.e. 

Taking into account expressions (13) and (15) , we obtain the following 
formula for the mean increase in plasma ion energy -rr = 2t, (R^f • 23) 

~ 2 

dw lH\ „ 

T^~kj'"^'- (16) 

In actuality, the energy is absorbed by resonance ions with V|| « 
ss "^res ~ T~' These ions may be regarded as magnetic dipoles with the mag- 
netic moment _ m,"i ^ Zl which moves in the wave magnetic field with a 

38 



strength which slowly changes In time and space. Within the frame of 
reference in which the wave is at rest, the equation of dipole motion 

ma==-V.^Il (17) 

dz 

leads to the law of energy conservation 



+ V-Hz = const. (-L3) 

0) 



Ions with velocities V|| in the rr-^r - Av < V|| < r™ + Av range, where 



Ay 



/^- 



(19) 



are trapped in the potential well and effectively interact with the wave. 
It is apparent that the number of resonance particles per unit volvmie is 



" ".Vl, (^o* 



Due to the shift of ions under the influence of the force -y —, 

dz 
the distribution function in the resonance region is distorted during the 
time T^^^^^^ , during which a trapped particle covers a distance on the 

order of the potential well width, i.e., 

^ei-~T7Air- (21) 

Relaxation of the distribution function due to collisions in a narrow re- 
gion of the width '\^ A v close to v|| = r— - 'vv-j^ occurs during the time period 



^.er'-dT- (^^> 



where 



"^l = .^7S^ 



4Y2Knae*A' 



/45 



It is apparent that the ion distribution function will be distorted by an 
insignificant amount if Tnonlin. ^^ ""^rel* ^^ critical value of the varia- 
ble magnetic field strength H = Hj,^., at which the distribution function dis- 
tortion becomes significant, is determined from the condition Tj^q^j]^ ^^^^ _ '\. 
-v Trel. Taking into account (19), (21), and (22), we obtain (Ref. 23) 






39 



Expression (23) for H^^ can be obtained by another method. An in- 
crease in the perpendicular velocity component (or a change in the mag- 
netic moment) in the presence of a variable magnetic field is determined 
in the drift approximation from the following equation 



Vj^o> dH, 



"- = 2^^ • &-• (24) 



The time of nonlinear distortion of the distribution function with 
respect to the velocity vj_ under the influence of the field Ea — i.e., 

under the influence of the force "^ — 5. in (24) — is determined according 

9z 
to the relationship 

Tnonlln - -j^^^' (25) 

where 






(26) 

If the ions obtain a velocity increase % Avj_, then the relaxation /46 
time of the velocity distribution function is determined by 



^r*X -'.(^) • (") 



The critical value of the magnetic field strength H = H^j-, at which 
great distortion of the ion distribution function f(v||, vj_) occurs close to 
V|| = T—, which is determined from the condition i^nonlin '^^rel' coincides 

with the value of (23) in the case of :; — 'vv^ . It is also apparent that the 

k|| 1 

distortion of the velocity distribution function of the particles close to 

V|| = T— is not significant, if the energy acquired by resonance particles 

■^ II 
having the velocity V|| (in the ~ - Av_|_ < v|| < '^ + Avj_ interval), during 

K-ll II 

the time between two collisions t^^ 



dw 
df 



^ ^'-^dr-STT^'-lTT:) ""Tixi, (28) 

ires -L 



is small as compared with their thermal energy T^. The critical field 

strength H, determined from condition — t^^ '^''^1' coincides with (23), 

"'- res 



40 



During heating with fields H = aH^j. under the condition r— o-vi (it is 
apparent that the fulfillment of this condition requires that the fre- 
quency be increased proportionally to / T^ as the temperature increases) 

^ _ "'^^ = °'^o (Toy- (29) 

dt T,(<oT.)'/. To(a>oT„)V.\^r,.j • 

where Tq , tq and wq are the initial values of Tg, t^ and to. 

Assuming that w ~ T^ and integrating equation (29), we obtain the 
following expression for the ion temperature (Ref. 23) 



'•'-•['+^Ti^r-4'-T^.r- 



Plasma heating by means of ion Cherenkov resonance may be intensified, /47 
if — instead of one wave — several waves are employed with phase veloci- 
ties differing by 2 - 3 Av, so that all particles in the -v^^ < v|| < v^ 

range may be resonance particles. In order to do this, it is sufficient to 
produce a wide wave packet with a phase velocity scattering of A(r~) ~ v^^. 






It is apparent that the total number of such waves is 7~'~1/ — ^. It is also 

evident that in this case the critical strength of the variable magnetic 
field responsible for great distortion of the ion distribution function over 

the entire |v| < v^ range is 



^cr.tot.~Ai;^cr~l^^cr^o /o^x.)'/. " (31) 



In the case of heating by the field H^ ~ "^cr tot ^^ ~ °'^^cr^ ^"^ ^^^ 
case of a wide wave packet, we have 

dw "l (h\ ^ «'^i 

-7/-~A5\F„j "'^■- ^- (32) 

Integrating expression (22) , we obtain 

Formulas (32) and (33) , which determine the ion temperature during the 
heating of plasma ions under optimum conditions, when practically all the 
plasma ions participate in absorption of the high frequency field energy, 
are similar to formulas (1) , (4) , and (7) , which determine Joule electron 

41 



»■ 
:* 



barely changes). One advantage of this method 



heating by subcritical "constant" fields. 

In the case of a = 0.01, no 'v lO^^ i/cm^, Tq = 100 ev, Hq % lO^G, 
and wq '^ 10^ sec"-^, we obtain: tq o- 10~^ sec, T^ = 50 kev during the time 
t '\. 1 sec, while in the case of t = H^ -vSOOG, Ej^ '\j 50 v/cm at the end 
of heating oj 'v 20 uq '\. 2-10^ sec"-^ '\' 0.2a)i and H^ % 20G, E^ 'h 40 v/cm 

(E^>^ /I^y/e E^ 

is the fact that energy is transmitted directly to the plasma ion com- 
ponent. 

In the case of a very non-isothermal plasma (Tg >> T^) , as was shown 

in (Ref. 24), a resonance relationship between the outer circuit and the /48 
plasma can exist in the case of V . « Vg. In this case for Tg < 10T-j_ ion 

absorption is still significant (on the order of electron absorption), and 
resonance Vj, ss Vg can exist. The energy absorbed at the maximum increases 

by 10 - 100 times as compared with (16) (Ref. 23). 

An expression for the high frequency power absorbed by a nonuniform 
plasma cylinder in the case of -:-— 'V/ v^ was obtained in (Ref. 25). 

As of the present , the effect of Cherenkov absorption of a magneto- 
sound wave by plasma ions has not been studied experimentally. 

Cherenkov Electron Heating By the Field of a 
Magnetosound Wave (Helicons) 

A rapid magnetosound wave can be propagated in a plasma with a large 
density {^l^^^^e , where Q^ — i / l!!f!^ — Langmuir frequency, Wg, = 2. — 

electron gyrofrequency) ; the refractive index of this wave in the a)i<<u)<<a)g 
frequency region is determined by the expression 

(in this frequency region, rapid magnetosound waves are called "whistling 
atmospherics", or simply "atmospherics", "whistles", and "spiral waves") . 



Cherenkov whistle absorption by plasma electrons is weak not only 



for -r— » Vg, when the damping coefficient is exponentially small (Ref . 26, 
■^11 

42 



27), but also for ^ < v^. In the last case (Ref. 28, 29), we have 



k^^v, 



(35) 



Since the whistle field readily penetrates a dense plasma, these 
whistles may be employed to heat the plasma electron component, while 
electrons located within the plasma cylinder will be heated up due to /49 
the comparatively small absorption coefficient (35) . 

If k|| a '\' 1, then — substituting (35) in equation (14) — we obtain 



^~(^W„j"'^4;7--^j • (36) 



Since the factor — • 2~ differs very little from unity in order 



of magnitude, this formula coincides with formula (16) (substituting Tg 
by T^) . However, since u >> co-j^ in the case under consideration, and since 
formula (16) was obtained for the frequencies u << co- , for the same values 

of H electron heating by the whistle field takes place much more rapidly 
than Cherenkov heating of ions by a low frequency field. 

The change in the electron distribution function with respect to vj_ 
by the quantity 



Ao 



-~^^1/ ^ 



*ii", (37) 



is insignificant, if the time of nonlinear distortion of the distribution 
function Tj^q^^j^^j^^ r^ — — - — is a little greater than the relaxation time 

/Avj_\ ^ 
T]-el '^''^el J • '^^ critical strength of a variable magnetic field 

H = Hj,j., at which the collisions cannot equalize the distribution function, 
is determined from the condition T^^onlin '^ ^rel* 

^^ (-.)'n"'j '~~(^'^'~w) ■ (^^) 

This expression for H^r is obtained under the condition that resonance 
electrons ^^ — Ay^<i;|, <-ii-+Au^j, which acquire the following energy 

43 



per unit of time 



dw 
dt 



res 



dw f, (h Y' _, U, STtrtoT-A*'' 

dt--^'-\Tu] '"^'l^'"^ ' (39) 



If H = H during the time "^ t^, acquire energy which is on the order of /50 
the thermal energy. 

During electron heating by whistle fields with H = cxHj,j., the energy 
acquired by one electron on the average per unit of time is 

dw '^'T, U Jjlj\'' 

dt T^(<ox^)'/'\<o/ 87:«„rJ • (40) 

Assuming, for purposes of simplicity, that (— . - — %-■] — 1, we obtain 

Under optimum conditions, when a wide wave packet is employed and when 

a large portion of electrons with a velocity of |v||| < v„ are resonance 

"^ r'"' 

electrons, electron heating by a field with H2 '^' aHj,^.^ ^^^'x, a v Hj,j.Ho is 

determined by the following expression 

dw a^r, ^ ^ / a'AV. 

These formulas describe Cherenkov electron heating by a whistle field under 
optimiffli conditions, if all the plasma electrons participate in energy absorp- 
tion. They are similar to formulas (4) and (7) , which may be used to deter- 
mine the electron temperature increase during ohmic heating by fields with 
subcritical strength. 

The study (Ref. 30) was devoted to a theoretical investigation of 
Cherenkov electron heating by the electric field of a rapid magnetosoxmd 
wave propagated in a nonuniform plasma cylinder. Cherenkov whistle ab- 
sorption was determined experimentally in (Ref. 31). 

Ion Cyclotron Resonance . 
Absorption of Alfygp Wave 

Ions are heated most effectively by the field of a high frequency wave 
under conditions of ion cyclotron resonance. If the frequency of a wave 

44 



propagated in a low-pressure plasma is close to the ion cyclotron fre- 
quency, then there will be a large number of particles having the velocity 
V||, which is close to the particle resonance velocity /51 



v^^s= 



res- A J . (43) 

and effectively Interacting with the wave field. Therefore, the wave 
energy absorption by ions will be large. Cyclotron damping of waves is 
the inverse of cyclotron radiation of waves by charged particles in a 
magnetic field. This effect was first studied in (Ref. 32), where they 
investigated the damping of magnetohydrodynamic waves propagated along the 
magnetic field. A study of the cyclotron absorption of waves having a 
frequency of u) w (u^^ was pursued in (Ref. 33) [see also the studies (Ref. 20, 
34, 35)]. 

As is well known, in the frequency region on the order of ojj^ in a 
cold plasma (T = 0) there are two branches of oscillations corresponding 
to an Alfven wave (which can only be propagated in the case of lo < o)^) 
and corresponding to a rapid magnetosound wave. As the Alfven wave fre- 
quency approaches Wj^, its refractive index and cyclotron damping coefficient 

increase. In the case of w^ - u < k|| v^, the propagation of this wave by 
strong damping is impossible: Re k ~ Im k ~ 1/6^, where 6^ is the depth to 
which the field penetrates the plasma (Ref. 28, 36, 37) 

A magnetosound wave is absorbed slightly in the case of |w — oj. | < k|| v.. 

Let us first examine cyclotron heating of a plasma cylinder by an 
Alfven wave which is strongly damped. The azimuthal currents (12) excite 
this wave if k|| 6^^ ~ 1. Also assiming that k|| a ~ 1 and w^ _ co < k|| v^, 

we obtained the following expression from formula (13) for the mean energy 
acquired by one ion per unit of time 



dr~8^'~\^/'"^/8^n„r,- (45) 



It thus follows that the electromagnetic field energy which is acctraiulated 
■q— I is absorbed during a period of time on the order of 

1 



In addition, a comparison of (45) and (16) shows that, for the same /52 
variable magnetic field strength, the energy absorbed by the plasma in 
the case of co ss Wj^ is considerably greater than in the case of Cherenkov 

45 



resonance. In the first place, this is due to the difference in the fre- 
quencies (■v^)~ <JJ and, in the second place, it is due to a large factor 



H§ 



in expression (32). On the other hand, for the same values of H 
SirnoT^ 

and k|| the electric field strength in the case of cyclotron resonance is 

times greater than in the case of Cherenkov resonance (w^^^^gj. — 

"^Cher. 

the wave frequency under Cherenkov resonance conditions) . 

In order to determine the number of resonance ions responsible for 
energy absorption during cyclotron resonance, let us Investigate the parti- 
cle motion in the field of a flat cyclotron wave : 



(46) 



Hx = Hsm{kiZ — wt), Hy = —Hcos(kiZ — iot); 

H H 

Ex = ——cos{ktZ — u)/). Ey = — —sin {kiZ — wt), 
"i "« 

where n\\ = — = ■ — is the refractive index. 

In the absence of a wave, the ions move along a spiral: 
V = Vo = {v°^ cos (u),/ + cpo), — v°^ sin (co,-/ + cpo). ^ i )• 

When there is a weak field, the velocity perturbation is determined 
according to the equations of motion 

"x= — J~ cos [k,Z — wt^-\- WiVy-, 

K = — ^nr- sin (kiZ — wt\ — WiV^; 

z - -^ (vlHy - vlH^ = ^ cos O, 

where O = A n 2 + (u)i — u)) ^ + <Po. 

Assuming that u = v + iv„, from the first two equations we obtain /53 

X J 

u 4- iwiu = -e \ ' '. 

'«<•" II 

Thus, assuming that u = uoe~-^"i , we obtain 



46 






The third equation can be written as follows 

» eu°Hki, 

<5 + i,^ " COS = 0. 



<^A„ ^ _ (47) 



Let us first investigate nonlinear distortion of the distribution func- 
tion close to v|| = V , caused by a variable magnetic field. The law of 

energy conservation follows from (47) 

i. <I)2 + _t 1 sin <D == const. 

The time of nonlinear distortion of the distribution function close to 
V|| = Vj-gg equals the potential well flight time of trapped particles 



having the velocity . ^ ^^, ,, _ „ 

" ' mkuc 






Tnonlin 



fe„AU| 



We obtain the following expression from the condition t^q^-^^j^j, ^ ^rel ~ 
for the critical value of the magnetic field strength 

^"^ {•^^iY'A Hi ) • (48) 



/ Avii \: 



This expression may also be obtained from the formula for the nonlinear 
decrement of cyclotron wave damping, which is determined on the basis 
of the quaslllnear theory (Ref . 38) in the case of v^^^ >> v^, if we set 



V 

res 



Vj^ and k|| ~ 7" in formula (25) of the study (Ref. 38) 



However, the influence of the accelerating field E on the distribu- /54 
tion function change in a plane perpendicular to Hg is considerably 

stronger than the influence of a variable magnetic field on the distribu- 
tion function change with respect to velocity along Hq . 

Let us determine the time of nonlinear distortion of the distribution 
function by the field E: 

1 



where 



^nonlin -- ^ , . , 



47 



^"^~°'/|-575;~"'/l(s4)"' <''' 

Equating t , . and t i '^ ""^ • I — —I ^^ the case of H -v H^,^, we obtain 

1 \Vi / 

"cr (a>T.)V.^ //2 j • (50) 

We obtain the same expression by assuming that only a group of resonance 
particles with v - Av_|_ < V|| < v^.^^ + Avj_ participates in energy absorp- 
tion. Resonance particles acquire the following energy per unit of time 



dw 
IF 



tes V^j '"'^'(^J • 

In the case of H '\. H^,^, during the time ^ t^^ these particles collect 

the energy '^ T^. The critical field strength (50) is considerably less 

than (48) . This means that the influence of nonlinear effects caused by 
the electric field is manifested for smaller field strengths than is the 
case for the influence of a variable magnetic field. 

^^ dw 

In the case of H = aH^^., we obtain the following expression for -r— 



dt ^.r,y/.\8nn,T, ,^ („,^)'A I 87.«„ro j [Ti j ' (5^) 



Thus, we have 



Ti-T, 



,+,_■■' f"o 



"■^0 {"^o)'''' \»^'hTi 






(52) 



Let us now present a numerical example. Let us set uq 'v 10 ^ cm~^, /55 
Ho ^ 5'103g, uij_ '^ 5*10'^ sec-^ and a -v 1. Then in the case of T^^ = 10 ev, 
we obtain: t^ -v. 3'10"^ sec, H '\. 2G and -^ "^ 300 kev/sec; if T^ 'V' 100 ev, 

then T^ '^10"'* sec, E'^ 0.4G and ^ '^ 30 kev/sec; if T^ 'x, 1 kev, then 

T4 '\j 3*10"^ sec, H 'v O.IG and -r-r ^ 3 kev/sec. 
-*- at 

In the case of Tq "^ 100 ev and H ^ H , the temperature T. ^ 10 kev 
is achieved during the time t '\/ 3 sec. 



48 



Cyclotron heating by fields with subcritical strengths may be in- 
tensified, if several waves (wave packet) are employed having phase 
velocities which differ with respect to 2 - 3 AvJ_. The number of such 

waves IS '^ —i. , and the critical strength of the total magnetic field is 
AvJ_ 



^cr. tot "^ ^ Hpr^O' where B.^^ is determined according to (50). In the 

dw 
case of heating by the field H '^' otH tot' ^^ this case for -rr we ob- 
tain (32), and for T^ we obtain (33). 

Cyclotron wave absorption in a plasma cylinder with a constant 
(over the cross section) plane and temperature was analyzed in (Ref . 33, 
39). The absorption of long wave oscillations in a nonuniform plasma 
cylinder was studied in (Ref. 40). 

In the case of a plasma with a large density and magnetic fields 
with a high strength, the skin depth S^ (44) is small. When 6^ is less 

than the plasma radius, plasma heating by an Alfven wave is ineffective, 
since the wave energy is absorbed only by ions located on the periphery 
of the plasma cylinder. 

In order to avoid this difficulty, Stlx (Ref. 41) employed the in- 
genious idea of "magnetic beaches". If the plasma cylinder is placed in 
the field Hq with a slowly decreasing strength, then cyclotron damping is 
exponentially small in the region w^ - w >> k|| v^^. An Alfven wave readily 

penetrates the plasma, and they may be excited resonantly. When the wave 
is propagated along Hg , the difference w. - oi decreases, while the refrac- 

tive index n|| '\- 1/ u. - to and the damping coefficient x— exp — o "o - /56 

^ [ 2* II t;f J 

increase. If the magnetic field decreases smoothly, the reflection co- 
efficient will be small, and the wave will be absorbed at the approach 
to the region of strong cyclotron damping, where w^^ - u^ k|| v- (region 

of the "magnetic beach"). The field behavior close to the magnetic beach 
was determined in (Ref. 42, 43). Absorption of Alfven waves at the magnetic 
beaches was determined experimentally in (Ref. 44). Experiments perfoinned 
at Princeton closely coincide with the theoretical computations of cyclotron 
damping in a linear approximation [see the simmiary in (Ref. 45)]. On the 
other hand, in several devices for ion cyclotron heating (see, for example, 
[Ref. 46]) the variable field strengths are on the order of, or even con- 
siderably larger than, the critical strengths. Under these conditions, 
we must expect a decrease in the field energy absorption, as compared with 
the case of weak fields. 



49 



Cyclotron Absorptio n of a Magnetosoun d Wav e 

The above-mentioned difficulty entailed in high frequencv heating 
of a plasma having a large density by a strongly damped Alfven wave is 
unimportant for a rapid magnetosound wave. A magnetosound wave is ab- 
sorbed in the case of o) « u^ to a considerably lesser extent, and can 

penetrate the plasma readily. The damping coefficinet of this wave in 
the case of [ai - (^j| ^ k]| v^^ is on the order of (Ref. 28) 






(53) 



c 

where nj_ '^ ~. 

^A 



If k|| a '^' 1, we obtain the following expression (Ref. 39) for the 

energy absorbed on the average by one ion per unit of time from the formula 
(13) , taking into account (53) 

Plasma heating by a magnetosound wave is greatly intensified, if the 
frequency to ;i; u^ coincides with the eigen oscillation frequency of a plasma 
cylinder w^^g. In this case, the field strength in the plasma increases /57 

in the case of loj_„„ - ail :^o)i/ ""° '" , as compared with the non- 
res y ^2 

resonance case, by a factor of 1 / ^o * 



SmiaTi ' 



It is apparent that we then have 




(56) 

Multiple resonance: u) = 2w£, may be employed to heat a plasma with a 
large density by a magnetosound wave. The damping coefficient of the mag- 
netosound wave in the case of |ci) - Zw^ | ^ k|| v-j^ is (Ref. 28) 

^_ /-8^/_Q^Y (57) 

«x V "" Vw'J 
where 2- = i /'^'^^""q is the ion Langmuir frequency. It follows 

y '"i 

50 



from (57) that in a plasma having a large density Qj_ > k|| c resonance 
at the multiple frequency w « 2a)j is more advantageous than at the main 
frequency . 

The heating of a plasma cylinder which is uniform across its cross 
section was examined for o) « nto^^ (n = 2,3, ...) in (Ref. 47), and the 
case of wavelengths (k 1 1 a << l)in a nonuniform plasma was investigated in 

(Ref. 48). Cyclotron absorption of magnetosound waves in the case of 
0) = 03^ and to = 2co. has not been determined experimentally as yet, although 
the inverse effect — cyclotron radiation of ions in a dense plasma — 
was studied recently (Ref. 49). 

The statements presented above Illustrate the following: 

(1) Cherenkov and cyclotron plasma heating with weak fields, when 
distortion of the ion distribution function is compensated by collisions, 
occur at the same rate as ohmic heating by a "constant" electric field, /58 
whose strength is less than the critical strength; 

(2) Plasma heating by weak fields up to thermonuclear temperatures 
(T^ ~ 50 kev) takes place over a long period of time. However, this time 

is less than the time for containing a plasma in a thermonuclear reactor 
with a positive balance; 

(3) When heating is performed with fields having subcritlcal strengths, 
it is more advantageous to employ a plasma with a great density, since col- 
lisions occur more frequently in it, the critical field strengths are larger, 

and the heating time is less than (t ~ — ) . 

no 



REFERENCES 

1. Transactions of the Second International Conference on the Peaceful 

Utilization of Atomic Energy. Reports of Soviet Scientists (Trudy II 
mezhdunarodnoy konferentsii po miromu ispol'zovaniyu atomnoy energii. 
Doklady sovetskikh uchenykh) . Gosatomizdat , Moscow, 1959. 

2. Transactions of the Second International Conference on the Peaceful 

Utilization of Atomic Energy. Selected Reports of Foreign Scientists. 
Physics of a Hot Plasma and Thermonuclear Reactions (Trudy II mezhdun- 
arodnoy konferentsii po mirnomu ispol'zovaniyu atomnoy energii. 
Izbrannyye doklady inostrannykh uchenykh. Fizika goryachey plazmy 
i termoyadernyye reaktsii) . Gosatomizdat, Moscow, 1959. 

3. Zavoyskiy, Ye. K. Atomnaya Energiya, 14, 57, 1963. 

4. Fanchenko, S. D. , Demidov, B. A., Yelagin, N. I., Ryutov, D. D. Zhurnal 

51 



Eksperimental'noy i Teoreticheskoy Fizlki, 46, 497, 1964. 

5. Chechkin, V. V., Vasil'yev, M. P., Grigor'yeva, L. I., Smerdov, B. I. 

Present Collection, 7. 

6. Sukhomlin, Ye. A., Suprunenko, V. A., Reva, N. I., Tolok, V. T. 

Present Collection, 118. 

7. Alexeff, B. Phys. Rev. Lett., 7, 196, 1963. 

8. Spittser, L. Physics of a Fully Ionized Plasma. XL, Moscow, 1957. 

9. Buneman, 0. Phys. Rev. Lett., 1, 104, 1958; Phys. Rev., 115, 503, 

1959. 

10. Suprunenko, V. A. Author's Abstract of Candidate's Dissertation 

(Avtoreferat kandidatskoy dissertatsii) , 1963. 

11. Suprunenko, V. A., Faynberg, Ya. B. , Tolok, V. T., Sukhomlin, Ye. A., 

Volkov, Ye. D. , Rudnev, N. I., Reva, N. I. Atomnaya Energiya, 14, 
349, 1963. 

12. Gordeyev, G. V. Zhurnal Eksperimental'noy i Teoreticheskoy Fiziki, 

27, 19, 1954. 

13. Drummond, J. E., Rosenbluth, M. N. Phys. Fl. , 5, 1507, 1962. 

14. Vedenov, A. A., Velikhov, Ye. P., Sagdeyev, R. Z. Yadernyy Sintez, 

1, 168, 1961. 

15. Vedenov, A. A. Atomnaya Energiya, 12, 2, 1962. 

16. Petviashvili, V. I. Author's Abstract of Candidate's Dissertation 

(Avtoreferat kandidatskoy dissertatsii), Tbilisi, 1963. 

17. Kadomtsev, B. B. , Petviashvili, V. I. Zhurnal Eksperimental'noy i 

Teoreticheskoy Fiziki, 42, 2234, 1962. 

18. Petviashvili, V. I. Zhurnal Eksperimental'noy i Teoreticheskoy 

Fiziki, 45, 1427, 1963. 

19. Lowson, J. D. Proc. Phys. Soc. , 70B, 6, 1957; in the Book: Controlled 

Thermonuclear Reactions (Upravlyayemyye termoyadern3rye reaktsii) . 
Gosatomizdat, Moscow, 12, 1960. 

20. Berger, J. M. , Newcomb, W. A., Dawson, J. M. , Frieman, E. A., Kulszud, 

R. M. , Lenard, A. in the Book: Transactions of the Second Inter- 
national Conference on the Peaceful Utilization of Atomic Energy. 
Selected Reports of Foreign Scientists. Physics of a Hot Plasma 



52 



and Thermonuclear Reactions (Trudy II mezhdunarodnoy konferentsii /59 
po mirnomu ispol'zovaniyu atomnoy energll. Izbrannyye doklady ino- 
strannykh uchenykh, Fizika goryachey plazmy i termoyadernyye reakt- 
sii). Gosatomlzdat, Moscow, 181, 1959. 

21. Sllln, V. P. Zhurnal Eksperimental'noy i Teoretlcheskoy Fizlkl, 23, 

649, 1952. 

22. Stepanov, K. N. Zhurnal Eksperimental'noy i Teoretlcheskoy Fizikl, 

34, 1292, 1958. 

23. Stepanov, K. N. Zhurnal Eksperimental'noy i Teoretlcheskoy Fizikl, 

45, 1196, 1963. 

24. Shapiro, V. D. In the Book: Plasma Physics and Problems of Controlled 

Thermonuclear Synthesis (Fizikl plazmy i problemy upravlyayemogo 
termoyadernogo slnteza) , 1. Izdatel'stvo AN USSR, Kiev, 62, 1962. 

25. Dolgopolov, V. v., Stepanov, K. N. Yadernyy Sintez, 3, 205, 1963. 

26. Shafranov, V. D. Zhurnal Eksperimental'noy i Teoretlcheskoy Fizikl, 

34, 1475, 1958. 

27. Gershman, B. N. Izvestlya Vuzov. Radiofizika, 1, 4-5, 49, 1958. 

28. Stepanov, K. N. Zhurnal Eksperimental'noy i Teoretlcheskoy Fiziki, 

38, 265, 1960. 

29. Stepanov, K. N. Izvestlya Vuzov. Radiofizika, 6, 403, 1963. 

30. Dolgopolov, V. v., Yermakov, A. I., Nazarov, N. I., Stepanov, K. N. , 

Tolok, V. T. Yadernyy Sintez, 3, 251, 1963, 

31. Nazarov, N. I., Yermakov, A. I., Dolgopolov, V. V., Stepanov, K. N. , 

Tolok, V. T. Yadernyy Sintez, 3, 255, 1963; Zhurnal Eksperimental'- 
noy i Teoretlcheskoy Fiziki, 45, 1260, 1963. 

32. Gershman, B. N. Zhurnal Eksperimental'noy 1 Teoretlcheskoy Fiziki, 

24, 453, 1953. 

33. Stix, T. H. Phys. Fl. , 1, 308, 1958. 

34. Shafranov, V. D. In the Book: Plasma Physics and Problems of Con- 

trolled Thermonuclear Reactions (Fizika plazmy i problemy upravlya- 
yemykh termoyadernykh reaktsiy) , 4. Izdatel'stvo AN SSSR, Moscow, 
426, 1958. 

35. Sagdeyev, R. Z. , Shafranov, V. D. In the Book: Plasma Physics and 

Problems of Controlled Thermonuclear Reactions (Fizika plazmy 1 



53 



problemy upravlyayemykh termoyadernykh reaktsiy) , 4. Izdatel'stvo 
AN SSSR, Moscow, 430, 1958. 

36. Stepanov, K. N. Zhurnal Eksperlmental'noy 1 Teoreticheskoy Fiziki, 

36, 1457, 1959. 

37. Doyle, P. H. , Neufeld, J. Phys. Fl. , 2, 39, 1959. 

38. Yakimenko, V. L. Zhurnal Eksperlmental'noy i Teoreticheskoy Fiziki, 

44, 1534, 1963. 

39. Vasil'yev, M. P., Dolgopolov, V. V., Grigor'yeva, L. I., Smerdov, 

B. I., Stepanov, K. N. , Chechkin, V. V. In the Book: Plasma 
Physics and Problems of Controlled Thermonuclear Synthesis 
(Fizika plazmy i problemy upravlyayemogo termoyadernogo sinteza) , 
3. Izdatel'stvo AN USSR, Kiev, 96, 1963; Zhurnal Teoreticheskoy 
Fiziki, 34, 974, 1964. 

40. Dolgopolov, V. v., Stepanov, K. N. Zhurnal Teoreticheskoy Fiziki, 

34, No. 7, 1964. 

41. Stix, T. H. Phys. Fl. , 3, 19, 1960. 

42. Stepanov, K. N. Yademyy Sintez, 3, 154, 1963. 

43. Piliya, A. D. Zhurnal Teoreticheskoy Fiziki, 34, 93, 1964. 

44. Hook, W. H. , Tenney, F. H. , Brennan, M. H., Hill, H. M. , Stix, T. H. 

Phys. Fl., 4, 1131, 1961. 

45. Hook, W. H. Nucl. Fusion, 4, 33, 1964. 

46. Nazarov, N. I., Yermakov, A. I., Tolok, V. T. Present Collection, 1. 

47. Dolgopolov, V. v., Stepanov, K. N. Zhurnal Teoreticheskoy Fiziki, 

33, 1196, 1964. 

48. Dolgopolov, V. v., Stepanov, K. N. In the Book: Plasma Physics and 

Problems of Controlled Thermonuclear Synthesis (Fizika plazmy i prob- 
lemy upravlyayemogo termoyadernogo sinteza), 4. Izdatel'stvo AN 
USSR, Kiev, 38, 1965. 

49. Golovin, I. N. Artemenkov, L. I., Bogdanov, G. F. , Panov, D. A., 

Plstunovich, V. I., Semashko, N. N. Uspekhi Fizicheskikh Nauk, 
73, 685, 1961. 



54 



DIELECTRIC CONSTANT OF A PLASMA. IN A DIRECT PINCH MAGNETIC 
FIELD AND IN A DIRECT HELICAL MAGNETIC FIELD 

V. F. Aleksin, V. I. Yashin 

Di electric Constant of a Plasma in a /60 

Direct Pinch Magnetic Field 

The study of electromagnetic oscillations in a nonuniform plasma 
has aroused a great deal of interest recently. This is primarily due to 
the discovery of several instabilities. A great ntomber of these studies* 
has been devoted to studying the electromagnetic properties of a slightly- 
nonuniform plasma located either in an almost uniform magnetic field with 
parallel force lines, or in an axially symmetrical magnetic field with 
helical magnetic force lines. 

In conjunction with these cases, when investigating the problem of 
controlled thermonuclear synthesis it is very important to study the plasma 
electromagnetic oscillations in pinch magnetic fields and in helical mag- 
netic fields and in a stellarator with helical current winding (Ref. 3), 
which has a more complex structure than magnetic force lines. 

As is well known, the electromagnetic properties of media are de- 
scribed by the dielectric constant tensor or the electroconductivity tensor 
related to it. In slightly nonuniform media, one can introduce quantities 
which are similar to the electroconductivity tensor and the dielectric 
constant tensor. In contrast to a uniform plasma, these tensors depend on 
spatial variables in wave vector space k and the frequency w. 

If E(k,(o) is the electric field strength, then the density of the 
current induced by this field can be written as follows 

ia{r, = Jc?kdu)o„p(k, w, r)£"p(k, w)e' i^'-""), (1) 

where a o(k, w, r) is the electroconductivity tensor of a nonuniform plasma, 

which is related to the dielectric constant tensor by the well-known rela- /61 
tionship 

Sap(k, 0), r) = 8ap + -;^a«p(k, o>, r). (2) 

The explicit form of the tensor o^^gCk, (o, r) may be found by different 



A detailed list of the literature is presented in the review articles 

of A. B. Mikhaylovskiy (Ref. 1), A. A. Rukhadze, and V. P. Silin (Ref. 2) 



55 



methods. In our case, it is advantageous to employ the method advanced by 
VJD. Shaf ranov (Ref. 4). In order to do this, it is necessary to determine 
the trajectory of the unperturbed particle motion and to solve the kinetic 
equation by the characteristics method. 

In the absence of a balanced electric field, the unperturbed charged 
particle motion may be described by the equations 

S=i™; §-v. (3) 

where B is the magnetic field strength in the plasma; e and m — particle 
charge and mass, respectively. In the cylindrical coordinate system, the 
strength components of a nonvortical, direct pinch magnetic field have the 
following form 

Br = Bo ^ §n sin az«; 



Bz = bJi+ J^fnCOsanzY, (4) 



where Bq is a uniform magnetic field; a = - — ; £„(!■) and g^C^) — functions 
of the coordinate r which are connected by the relationships 

"y^ no.gr,- i^=rno.U. (5) 

In the presence of a plasma, the field components, which are related to 
the pressure gradient and the longitudinal current and which may be found 
from the equations of plasma equilibrium, must be added to the nonvortical 
field (4). 

A solution of equations (3) -i-n the general case of arbitrary fields /62 
entails considerable difficulties. Let us examine the case (which is 



(a=U^«i) 



of practical interest) of a low-pressure plasma (^ = ^^z "^"^ 1) without a 

longitudinal current, in a pinch magnetic field (4) having a large uniform 
component Bo(fj^~ gn~ ^ << 1) • 

In solving nonlinear equations (3) , the presence of a small parameter 
6 makes it possible to employ — along with the drift approximation — the 
method of averaging (Ref. 5) when solving drift equations for flying parti- 
cles. The number of particles which are blocked is small, and their contri- 
bution to the tensor a^o may be disregarded. Avoiding cumbersome computa- 
tions, let us derive the final result of solving the equations of motion (3) 

within an accuracy of terms on the order of — ( *^r ~ ^"^ ) inclusively 



a)g 



y B mc y 



56 



r{t) =r — Yi ^[cosnci.{v ^t + z)—cosna.z]~ 
~;;r:t^'"(T — "'sO — sinTl; 

? W = ? + ^ + 2j^^^77i;^S-«[sin ««(«„/ + z) _ 

n=l 

— Sin naz] + ^— - [cos (y — wbO — cos 7]; 

00 

- ^ 2 [sin na (w n if + z) — sin naz]; 
2nav I, 

" U (^) "" " U ~ Zj 2^ /n [^°s «a (0 ^ + z) — COS naz]; 

n=l 

v±{t) = fx +2j 2""-L/"I^os«a(y / + z) —cos naz]. 



(6) 



n=l 



where v|| and v_[_ are the velocity components which are longitudinal and /^3 

transverse to the field, respectively; y — the initial phase of particle 
rotation around the center of a Larmor circle; v"^ — the averaged velocity 

component related to the drift of the Larmor circle center: 



(7) 



ro>gu2 



For purposes of simplicity, we have omitted the index for quantities 
taken at the initial moment of time t = in formulas (6) and (7). We 
shall also do this from this point on. 

By solving the kinetic equation, with no allowance for close collisions, 
by the method of characteristics, we can find the expression for the electro- 
conductivity tensor in a cylindrical coordinate system from the equation 
for the induced current density 

57 



I I IMI nil ^■■^IHI^i^HHaHII^H^^^BH^H^^^I^I^^H^^HmmHinillllll ■■■■■■■■■ 



where Fg is the equilibrium distribution function of the particles; k — 



m 



the wave vector with the components k^, k^, kj, (kj, = — , m — whole numbers); 

E — the sum with respect to ions and electrons. The electric field may 
i,e 



■>• 



be determined by the expression 

E (r, U)) = S I dkrdk, E {K, k., m) d'^rr^rik.'^tm, _ ^^ ^ 

For a slightly nonuniform plasma in a strong magnetic field, we may 
write 



^0 = (i + ^y^ ^^^)f{v„ ... w), (10) ^^ 



for the equilibrium distribution function within an accuracy of terms on 

vp 
the first order of smallness with respect to the parameter (vp — 

thermal particle velocity, a — characteristic dimension of the nonuni- 
form! ty for the main plasma state). Here, x — unit vector in the field 
direction; f — arbitrary function of the velocities v||, vj_ and of the 

variable ¥ which is an integral of the drift equations. Within an accuracy 
of the terms — , the quantity W coincides with the integral of the force 
line equations 

^=./-^ + 2r2|cos«az. (H) 

Due to the smallness of the parameter 6, we may approximately compute 
the function f which depends only on the coordinate r. We should note that,in 
order to avoid this, in a plasma with the selected distribution function 

div j =i^ we may add the component vB x 1 ^5^ " "^^^^ ^° ^^^ function Fq , 

where Integration is performed along the magnetic force line. However, as 
may be readily seen, in our approximation this term is small. 

Let us select the Maxwell distribution of particles with a nonuniform 
equilibrium density ng and the temperature T, which depend on 'P, as f : 



, = „.(,-fJ%(-^i^). 



(12) 



Retaining the important terms in (8) , after simple transformations we ob- 
tain 

i.c ^ ' (13) 



58 



where the following notation is employed: 



1=1 s, s' 






^n = arctg — 



(14) /65 



2jr 

^"P = —2!^ J ^T^a J yp (0 ^ (0 d^; 

^ (/) = exp {— / (o) — tiasv i—k^v^ — k^v,, )t--ax 
X [sin(-)f — co£/ — (j;)_ sin (y — <}))]}; 

The velocity components which are included in the expression for the ten- 
sor Qjjjg in a cylindrical coordinate system have the following form 

00 

"r{t) = V :^ COS {-i — iost) + 1] g^v , sitina {v it + z); 
r-^'=€'xSm(T-«.BO+ 1 2^^ V .cosna(^.^ + 2) + cr,; (15) 

°° 2 

^^ (0 = y I — X^ /" COS na (y , / + 2). 
By employing them, we can obtain the explicit form of the tensor Q g 

where the vectors q and q have the following form 

qr = Ux [l^p (?) cos if + iJ'p iS) sin ^\ + Vi~Jp Q.)\ 
9f = ^x [f -^p (^) sin <J; — iVp iS) cos ijjj + F/p {^);.' 



59 



gr=^v.[^Jp (S) COS t - iJ'p (^) Sin ^) 4°' - 4 I, , /, (^) S ^/4'V'«; (^7) _£66_ 

9^ = 0^ [|- /p (S) sin <{. + f-y; (S) cos <i<]4°> + Vp(04°'; 



cJJ' =: (u) — pcofi -^ nasu I — lav , — fe^u, — k^v j )~' 
(J (5) is the Bessel function; the sum with respect to 1 is taken from 
-00 to +00, and it is thus assxamed that g_n =-gj,, f. = jj, and f q = 0) . 

As is well known, the electroconductivity tensor may be employed 
to obtain the expression for the polarizability vector x which charac- 
terizes the density of the induced charge p , 

p = y dlcdoje-- C^—O X,E^ (k, w). (18) 

The components of the polarizability vector have the following form 

Za = ir. 2 e^ J ^y ^ dvlML ^f(v,, v\, W) $] h (^) 9- (19) 

I, t p=s — oo 

The expressions for the electroconductivity tensor and the polariza- 
bility vector may be simplified in the case of electromagnetic oscilla- 
tions, whose wavelength is much greater than the pinch modulation depth 
(rijj << 1). In this case, we must get rid of complex svmis and products, 

since the terms with s = s' =0 and the operator M = 1 will make the 
main contribution. 

Plasma Dielectric Constant in a Dielectric Helical 
Magnetic Field of a Stellarator 

The strength components of a nonvortical magnetic field from a helical 
current winding with the finite step L have the following form (Ref . 6) 



5r=5„ 2^„sin«9; ^^ = ^0 S^/'.cosnO; (20) J67_ 



n=l 



Bg = Bq — o.rB^, 



where Bq is the strength of a viniform, longitudinal magnetic field; 

2ir 
6 = (J) - az; a = ~7~', gn(^) ^'^'^ ^n(''^) — functions of the coordinate r, 

which are related by the following relationships 



60 



Just as in the case of a pinch, field, when deriving the explicit 
form of the electroconductivlty tensor, we investigated a plasma with a 
small gasokinetic pressure (6 << 1), without a longitudinal current, in 
a helical magnetic field having a large axial component 

Within an accuracy of terms on the order of — , the particle tra- 
jectories and velocities may be described by the following expressions 

— ("bO — sin y]; 
X [COS (y — iOBt) — COS yI; 

2 " - 

z(O=z+^.^ + 2^i;'J?[sin«(G+^)-sin/z0]; (22) 

f/- (0 = y-L cos (y — mst) + Vi YiSnS^mnl^ -{-—A; 
/•M2 = y^ sin (y — (BflO + y I 2/« cos n ^G + ^j + y^; 

o ~ — 

Vz (0 = ^^ — 2^ «"■ 2j /" COS «. [9 + —j. 



n=l 



where v., v^ and Vg = va - arv^ are the averaged velocity components re- /68 

lated to the motion of the center of a Larmor circle along the averaged 
force line and drift in a nonuniform field: 



(23) 



y = artv 1 -1 — Ml + — '- «2 -I ^ "s". 

fz = C, - -4 t> II 2j (/" + g") - ^ r'ao.3 "3- 

Here x is the torsion angle of the magnetic force lines: 



n=l 



61 



"1 == I S (/« + si - 2nfngn - 3aV^«/„^„); 

n 

„, = I- j; {fl +gl- 2nfngn - aV^n/„g„) + '-^; (25) 

n 
«3 = — g- 1] a^r^nfngn- 

n 

The distribution function Fq in a slightly nonuniform plasma may also 
be determined by (10) , in which 'F describes the equation of magnetic sur- 
faces: 

VF = r2 — 2rS^cosrt0. (26) 

n 

We again select the fimction f as a Maxwell function, with the density and 
temperature dependent on the magnetic surfaces V. 

As a result of the computations, we obtained the following from the 
general expression for o^^ (8) 



where 



..^=^i^Y,^'\ML^f{v,,v„^)Y, q^q^dv^dvl, (27) 



A A 
ML: 

I. I ' P=- 



M = n 2j ^sW^'(%)e ^'. 

n=\ s, s' 

<7. = y X [f ^p (^) cos ^ + iV; ($) sin <!>] + J^ y , /p {%); 

<7, = t; X [f ^P (^) sin t - fV'p (?) cos t] + ^./p (S) + g-^ " 1 ^p (^); 

^, = t; X [f ^p (^) cos i> — i7'p (5) sin <1^] c^ — ^ ^ » /p (^) X 



(28) /69 



62 



^, = f X (f Jp(i) sin <}. + iJ'p (?) cos t) c<,°> + v,Jp (E) 4°> + 



The quantities kx, kj_, ij;, g and the operator L are determined by the ex- 
pressions (14) ; it is assumed that g_ = -g , -f = f and f q = in the 

Jl^ A/ Xi 10 

sums over the index Jl . Correspondingly, the plasma polarizability vector 
in a helical magnetic field is 

A A 1 

p= 



X^a = w^e^ j ML j7/(o,, Ux, 'P) S Jp{i)qJVidv].. ^29) 



Just as in a pinch field, expressions (27) and (29) may be simpli- /70 
field, if the oscillation wave length is much greater than the difference 
between the maximum and minimum radii of the magnetic surface (n << 1) . 

We may employ the expressions obtained for o^o and x„ to study the electro- 
magnetic oscillations and plasma stability in pinch magnetic fields and 
in helical magnetic fields. This will be the subject of future research. 



REFERENCES 

1. Mikhaylovskiy , A. B. In the Book: Problems of Plasma Theory (Voprosy 

teorii plazmy) , 3. Gosatomizdat , Moscow, 141, 1963. 

2. Rukhadze, A. A., Silin, V. P. Uspekhi Fizicheskikh Nauk, 82, 499, 1964. 

3. Spitzer, L. In the Book: Physics of a Hot Plasma and Thermonuclear 

Reactions, 1. Atomizdat, Moscow, 505, 1959. 

4. Shafranov, V. D. In the Book: Problems of Plasma Theory (Voprosy 

teorii plazmy), 3. Gosatomizdat, Moscow, 3, 1963. 

5. Bogolyubov, N. N. , Mitropol'skiy, Yu. A. Asjnnptotic Methods in the 

Theory of Nonlinear Oscillations (Asimptoticheskiye metody v teorii 
nelineynykh kolebaniy) . Fizmatgiz, Moscow, 1958. 

6. Morozov, A. I., Solov'yev, L. S. Zhumal Teoreticheskoy Fiziki, 30, 

271, 1960. 

63 



SECTION II 
LINEAR PLASMAJ)SCILLA TIONS /71 

KINETIC THEORY OF ELECTROMAGNETIC WAVES IN A CONFINED PLASMA 

A. N. Kondratenko 

Problems of electromagnetic wave propagation in a confined plasma 
are of considerable interest to studies on methods of plasma heating, 
acceleration of charged particles, plasma diagnostics, and other possi- 
ble applications. The hydrodynamic theory of plasma wave guides for slow 
waves has been studied in great detail [see, for example, the articles 
(Ref. 1 - 4)]. However, the hydrodynamic theory does not encompass the 
important phenomena related to the particle thermal motion — for example, 
wave damping which is particularly great at small phase velocities. 
Since the phase velocity of a propagated wave V^ is less in the wave 

guides of slow waves than the speed of light, and since it may be compara- 
ble to the mean thermal velocity of electrons Vj^ or ions v>pj, the neces- 
sity of a kinetic examination becomes readily apparent. 

On the other hand, the confinement of a plasma leads to a new type 111 
of wave — surface waves — whose damping, as was shown in (Ref. 5, 6), 
is proportional to the thermal velocity of plasma electrons for v^jg << V$. 

In a nonconfined plasma, where there is no surface wave, the damping of 
the longitudinal three-dimensional wave is exponentially small (Ref. 7). 

Foirmulation of the Problem 

Let us investigate the propagation of slow electromagnetic waves in 
a plasma layer which is 2a thick in one direction, and is not confined in 
the two other directions. As is known (Ref. 4), waves propagated under 
these conditions are surface waves when there is no magnetic field. 

A self-consistent system of equations describing these processes con- 
sists of the Maxwell equations 

™tE==-^f: (1) 

-tH=|.f + ^- (2) 

and a linearized kinetic equation for the deviation of f^^ from the equill- 

briiam distribution function fo„ of the a type of particles (a = i — ions; 
a = e — electrons) , in which we shall disregard particle pair collisions 



64 



t+'J + ^(^ + }l«."l)^--0. (3, 



The standard notation is employed in equations (1) - (3) . We shall select 
the coordinate axes so that the z axis coincides with the direction of 
the wave propagation, and the x axis is perpendicular to the layer. The 
YZ plane is the plane of symmetry. 

Equations (1) - (3) must be supplemented by the boundary conditions. 
Let us assume that f„ = fj + f^, where fj is the distribution function 

for v^ > 0, and f~ — for v,, < 0. We shall select the conditions of the 

mirror image (Ref. 8) (the final result does not depend quantitatively 

on the reflection condition) as the boundary conditions for the function f * 

}t{ + a,V:c>0,v,)=fri±a,V;,<0,v,). (4) 

We obtain the boundary conditions for the fields from the Maxwell equa- /73 
tions (1) and (2) , integrating them over an infinitely thin layer which 
encompasses the plasma-vacuum boundary: 

Dispersion Equation 

We can write the dependence of the distribution function fa on 
time and the cordinate z for the fielctein the form exp i(k3Z - cot). If 

fOQ( is the function of energy, we obtain the following from equation (3) 

with allowance for the boundary conditions 

— a 
J^^e^n^^A 0^^^ ^ (6) 

— a 

_/•£.($) 3^ sin T (a -S)|, 



where 



fejV^— 0> 



In order to compute the currents j^, and j^, it is necessary to deter- 
mine the simi and the difference ft, + fl^ Employing the values (6), we 
obtain 



65 



a 

— a 

— a 

a 

2ie^ a/„, p (8) 

— a 

where ... ^ , . x ► /74 
„ , „ ^ 1 f cos T (S — a) cos 7 (X + a), ;c < ^, 

sin27a\(,Qs^(;j._Q,)(,os7(S + a), a:> $; (9) 

/( (x ^)^ ^ fsin7(^ — a)sin-r(A: + a), ^ < ^, 
""^ • ' sin2-jff\sin^(;c — G)sinf (^+a), a:>5. (10) 



The kernels K^ and K2 , which are expanded in Fourier series , have 
the following form 

'cos o„S cos a„ X 



Y' ''^°^ "n^ '^OS "n * 

Ki (x, i) = lL —Tir- ' (11) 



n=0 



_, sin a^S sin a^ x 
/C2(^, = ^2j f^-a^""' (12) 

where a = — ; the prime over the sum indicates that the sum term corres- 
n a -1 

ponding to n = must be multiplied by y. 

Employing the values (7) , (8) , (11) and (12) and using equation (1) 
to change from the fields E^, E^. to the fields E^, Hy, we may obtain the 

formulas for determining the currents : 

;^ sin a J ". ,.,v2.^f^ }j^^(dh. 7 ^M, 

+ 'P J "y sin «"W^ 2j ;^ J "'"^ J f - «r 5^; ' (13) 



66 



. '^' COS a„ ^ h A 2ie\ f v,dv^ f dv^ f a/o, 

n=0 \ a=i — ~ 



a=/ — oo 



where 



ckc 



The solution of the Maxwell equations (1) and (2) for determining /75 
the fields E^ and H^ leads to the following equations 

f-/^(l-p^)/f, = l^/.; (15) 

We may write the solution of the integro-differential system of 
equations (13) - (16) in the following form 

EzU) = ^'fan cos a„A:; Hy{x) = '^HynS'manX. (17) 

We obtained the following value for the Fourier components E^.^^: 



£.„ = _(_-!)« I //,(a)M+iML^\ (18) 



where 



A= J--/1 



(19) 



[i~ - A,) [a, + /| (1 - n] + i^B - ««) (^2 + ^n); 



0=t — oo 






67 



I I I II I I II iiiiiiniHiaiM ■■■iini i ■ 



Here fi^ = " ; pg — equilibritmi ion density which equals the 

c m 
■"a 

equilibrium electron density; f is the density normalized to unity 

(/f dV = 1) . 

oa 

Selecting Maxwell distributions as the equilibrium distribution /76 
functions f , we may write the equation for the plasma impedance 



" n=o 



1-PIi-SSQ'^ 



(21) 



We thus have 






i-PV-i]S<3- 



a={ 



V^Q^ 



a=l" 

Qi. = I a. J ^2 (a. - bz) e-^' [(a, - 6z)^ /„ (z) - K^]; 
Q2a = 1 0. f dzz^ (a, — bz) er^'Ia (z); 
Qa. = ?- a,6 J dz2 (a, - fcz)« e-^7a (z) 



(22) 



(23) 



; wt 






the mean thermal velocity of a type of particles ; 



b=^; h{z)= f t^ ) 

Equating the plasma impedance (21) to the vacutmi impedance 



^z(«) _. c ^ 
iHJa)- 0. '^s^^ 



2\2 



-ff. 



(24) 



we obtain the dispersion equation 



68 



J*3 2ja 



i-PMi-LipQ 



n=0 



s 



Q^ 



!u 



= (1 - P«)2 



(25) 



The sum (25) consists of the real and imaginary parts, since each Q con- 
tains the real part which equals the main value, and the imaginary part 
which is proportional to the residue of the integral I (z) . 



/77 



High Frequency Oscillations 
Let us set Vrt = 0, — -' <C Ir wi/ ->- oo. 
We shall employ the following notation 






Pa = ?^' Im 



n=0 *- ^ 



Qi 



(26) 



(27) 



Since a^^ are evenly included in the sum (26) , we may change from the 
sum to the integral (Ref . 8) 



^^=^^J^[^-p^(i-S'^^('^) 



dq. 



(28) 



in which integration over the contour C is performed from -«> to -H=°, passing 
around the singular points q = — from above. Let us deform the integration 

3. 

contour, and let us enclose it in the upper half-plane. Let us set 
qo = ikj_ — the value of q at which A(q) =0. We then have 






(29) 



since the integrals Qj have the following principal values Oj (j =1, 2, 3) 
when the electron pressure is disregarded: 

When computing the sum P2, we should note that — since a = 

= b T >> b — for small n, just as for a >> 1, the integral residue 

ICgV-pg e 

I(z) is exponentially small. Therefore, components with n which is 



69 



larger than a certain ng (i.e., shortwave components of Fourier expan- /78 
sion) , which may be determined from the condition o^ "^ 1, make the main 
contribution to the sum P2. For such n, b^ << 1, and Qj have the 
following values (we shall omit the index e) : 



Q^ = 20^ [ -1 + a7»= \2]e''dt - iV^ 
Q2 = a^°' (2 ] e''dt ~ i yii) ; 







(30) 



Q,=2ob' 



_a + U — i) e-o= (2 ]e''dt - i V' 



Since Qi '\' a^, Q2 "^ o, -^ '^ a, in the case of b^ << a^ , b^ << 1 for A, 

b 

which is determined by equation (22) , we obtain the following expression 

A = -4(i-|q,). (31) 

It t 

Let us employ the following notation: Qi = ReQi, Qi = Im Qi. We 

then have 



P2 = 






a^Q, 



1-J«> +bl« 



2 • 



(32) 



n > /!„ 



Since ng >> 1, we may change from the sum to the integral 



Si k,v 



^'~~Y^"^' 



e "s^Tc 



^--^Q\]+i^iQ\ 



(33) 



We may disregard the quantity Q^ in the denominator of the integral, 
and for Qi we may employ its value in the case of a '^ 1. In addition, 
making an exponentially small error, we may multiply the lower integration 
limit by <=°. We finally obtain 



P2=. 



l_e *3"t 



(O3 



e=i--^Q:/,-... 



The dispersion equation (25) now assimies the following form 



(l^zl£cihk.a + 



. 2 1 



^3f„ 



Yii 



(1 - n 



1 

i\2 



(34) 



(35) 



Let us assume that ks ->- ks = ks + 16 , 6 << ks. Then the real part /79 
of equation (35) produces a relationship between the phase velocity g and 



70 



the frequency of the propagated wave w: 

I 

and the imaginary part produces damping: 

S = -?= . ^3"t< _ (1— 32)(1— e) ^ 

Yti ' « 'p2e2(1 — e) , Mik^a'l^' (37) 

Formula (37) can be considerably simplified in the case of large 
(kj_ a >> 1) and small (kj_ a << 1) layer thickness. We have excluded 
3^ from equation (36) , and shall substitute it in (37) . In the case of 
k_|_ a >> 1, we obtain 

V~^' <- '{\-\^\f'l^' (38) 

where k = — . The damping determined by this formula coincides with the 

damping found in (Ref. 5, 6) for a reflection coefficient of P = 1. In 
the case of kj_ a << 1, we have 

8 = -f. kv.e 1^^ • (^' + ^W? . (39) 

It may thus be seen that wave damping is increased when there is a 
decrease in the plasma layer thickness. 

Ion-Sound Waves (v^^g >> V^ >> 'Vr^^) 



The dispersion equation (25) may be considerably simplified, if we 
set the speed of light c = 0°. In this case, we have 

A = -^dl+i l-T,§iQi + b'Q2+2Q,)A. (40) 

I L a={ -if 

A A 

We shall employ the following notation: A = Re Oj A' = I™ v2" /80 

1C5 -kg 

Then the real P3 and the imaginary Pi^ components in the sum (25) have 
the following form 

p - 2 y' A' . 



n=0 



4"* + A'" (41) 



^* = i2x^- (^2) 



n=0 



71 



We shall employ the same procedure in computing the sum P3 as for 
computing Pi, but we shall take the fact into account that in the case 

of ■; << 1 we have 



kgVTe 



We obtain 



Re FQi {q) + i Q2 (9) + 2Q3 (<7) I = -2 -^ ■ 

t Je " "re 



^3=^ 



(43) 



(44) 



where SS^ Q? m, 

x^ = ^Hl+?); | = ^^:3,. = 1-J; , = ^. 

Employing the integrals (23) , we find 

Im (Q, + i^Q^ + 2Q3). = - -?— ^ exp f - ^ . 

(1 + b^r^ ^ 

Since ^ , , 2 '^'^ ^ fo^ ^'^y '^j we then have 



(45) 



a — f . — ^ 



(1+6^)2 



IfeF-^'n^fe)"} 



Consequently, both electrons and ions make a contribution to the wave 
damping. The ion component P^. of the sum Ptj has the same form as P2 with 

a replacement of the indices e ->- i : 

In computing the electron component Pi^ of the sum P,^, and in computing /81 
P. ., we shall set A"^ » A'^. We then have 



4 yT 1 — 5< 1 









«-v^ + ^+. 






(48) 



Thus, the dispersion equation (25) has the following form for ion- 
sound waves 

^•^^ + /(P« + P.) = 1. (,9) 

We may thus find the expression for the spatial damping decrement 



72 



&, 



0-A:3(l+5) T;^— • (50) 



m- 



sh^xa 



In a nonconflned plasma, where there Is no surface wave, the damping of a 
three-dimensional ion-sound wave is determined by Cherenkov absorption of the 
wave energy by plasma electrons, and the ion contribution to the damping 
is exponentially small. As may be seen from formulas (47) - (50) , plasma 
ions also make a great contribution to the damping for a surface ion-sound 
wave. The physics here is the same as for high frequency oscillations — 
the Cherenkov absorption of the wave energy by plasma ions is particularly 
significant for the short wave components of Fourier expansion of the 
surface wave. 

Let us study equations (49) and (50) in special, different cases. If 
1 /p 
k3a(l + O » 1, then we obtain from equation (49) 

b2-^.^^; (51) 



3 y2 ,_2 
"re I 



4 



/i ■ v^^. 






8 =-h 

vL |8,|ll-e,- )-| <" e,- 4p.«; ^ • J 

kaa 

2lT 



(52) 



since for large ksa the sum included in Pj^^ approximately equals -J— 



It can be seen from formula (52) that the ions make a particularly /82 

1/2 
large contribution to the damping, if e. -^ -1. If k3a(l + ?) << 1, 

then ^ , „ ' 



^[-(■-'fT] 



(53) 



^ = ks\^^{P,i + PA>). (54) 

The following limiting cases are possible: (1) At a large plasma 
electron temperature or a small layer thickness, when — Se,— ^ < 1 (| ^| <C 1), 
the damping decrement is 



8: 



9 




ta 


e,- ^ jt<o 


V-. 


—2 ' 3 



a' I e, l^j 



(55) 



(in the case of ksa « 1 the sum included in Pi^^ approximately equals y) '> 
(2) For a large plasma layer thickness or an electron temperature which is 
not too high when _ Se, rL_ ;^ l (^ ~ _ 1), 



73 



!t 



-y ^ ^J .~[^-ef+/-8.Q3j (56) 



We cannot assume that a -»■ <» here, due to the condition of the initial ex- 
pansion. 

As may be seen from formulas (52), (55) and (56), the damping decre- 
ments of a surface ion-sound wave are large in different cases. Plasma 
ions make a significant contribution, and sometimes the main contribution, 
to the damping. The wave phase velocity is decreased with a decrease in 
the layer thickness ; therefore , the ion contribution to the damping in- 
creases, while the electron contribution decreases. 



REFERENCES 

1. Shuman, W. 0. Z. Phys . , 128, 629, 1950. 

2. Fainberg, Ya. B. CERN Symposium, 1, 84, 1956. 

3. Pyatigorskiy , L. M. Uchenyye zapiski Khar'kovskogo Universiteta, 49, 

38, 1953. 

4. Fajmberg, Ya. B. , Gorbatenko, M. F. Zhurnal Teoreticheskoy Fiziki, /83 

29, 549, 1959. 

5. Gorbatenko, M. F. , Kurilko, V. I. Zhurnal Teoreticheskoy Fiziki, 

34, 6, 1964. 

6. Romanov, Yu. A. Radloflzika, 7, 242, 1964. 

7. Landau, L. D. Zhurnal Eksperimental'noy i Teoreticheskoy Fiziki, 

16, 574, 1946. 

8. Aleksin, V. F. In the Book: Plasma Physics and Problems of Controlled 

Thermonuclear Synthesis (Fizika Plazmy i problemy upravlyayemogo 
termoyadernogo sinteza) , 4. Izdatel'stvo USSR, Kiev, 1965. 



74 



KINETIC THEORY OF A SURFACE WAVE IN A PLASMA WAVE GUIDE 



M. F. Gorbatenko, V. I. Kurilko 

As was shown in (Ref. 1), when a surface wave is propagated along a 
plane boundary of half-space occupied by a plasma having a temperature 
which is different from zero, Cherenkov absorption of the wave energy by 
thermal plasma electrons takes place. This leads to damping of this wave 
even when there are no collisions. In contrast to the damping of a longi- 
tudinal wave in an unconfined plasma, in this case for small thermal 
velocities the damping coefficient is proportional to the thermal velocity 
of plasma electrons. 

We investigated this phenomenon for the case of a plasma wave guide 
produced by a plasma layer having a finite thickness (21, |x| < 1). 
Since the electrons moving at a thermal velocity are successively reflected 
from both walls of the wave guide, it was not known previously that in 
this case the absorption investigated in (Ref. 1) will not decrease con- 
siderably. 

The initial system of equations consists of the kinetic equation for 
a high frequency addition to the distribution function and of a Maxwell 
equation: 

df 



aT + ^vr/-^^vr/o = o 



dt 



rot^ = l!57 + l.p 
c -^ ^ c dt 



TOtE 



c ' dt 
J= — e J vfdv 



(1) 



It is assijmed that the equilibrium distribution function is a Max- /84 



well distribution: f^ = —. 



exp 



vl + v^^ 



(the z axis passes along the 



direction in which a surface wave is propagated) . 

Let us write the solution for system (1) in the following form 
f{x, z, t) =f (x) exp [i {kaZ — wt)], E {x, z, t) = 

— E {x) exp [i (fegZ — 0)01- 

We then obtain the following equation for the distribution functions of 
particles moving in positive (+) and negative (~) directions along the 
X axis 



75 



»± 9»f. ^ .- (2) 



• /u)*<p± ± vj-l- + ?^ (y.e. ± «,£,) = D. 



We shall employ the following expression as the boundary conditions for 
the distribution function (|)* (x) in the planes x = + 1 of a plasma layer 

c?+ (x = — /) = pf~{x = — /), 
cp- (a: = /) = pf'^ (x = /)• 



„+,._A (0<P<1) ^^^ 



The solution for equation (2) which satisfies the boundary conditions 
has the following form 

+ P ^ {v,E^ — v^E^) exp r^ {x' + X)] dx' — p^ f {v^E^ + v^E^) x 

Xexpl-^{x'-x-2[)]dx'\; (4) 

'P"^^^ = ;^A { I ^"^^^ - '^^^^ ^^P [^ (^' - ^ - 20] dx' - 

- p j" (y,E. + t;;,£^) exp F- J^*(a;'+;c)1 dAT' - p^ [ (v^E,- v,E,) x 
X exp l'^{x' —x + 2/)l djc' 1 , 
where /85 

no — plasma electron density. 

By substituting system (4) in the expression for the current, and 
by substituting the latter in the Maxwell equation, we obtain a system of 
two integro-differential equations for the E^ and E^^ fields. In the 
general case, this system is cumbersome, and it is difficult to obtain an 
analytical solution for it. However, in the case of slight thermal 
scatter (v,j, -^ 0) , a solution may be found for the system. In actuality, 
in this case, as may be readily shown, current corrections due to the 
electron thermal motion decrease with an increase in the distance from 

76 



the surface of the plasma layer proportionally to 

exp[-i?v.|i_^|'A]^^=^^, 7j=^]. 

Since there are significant thermal scatter phenomena only in a layer 
having a thickness on the order of _Z. close to the plasma surface, they 

may be taken into account by means of effective conditions which the 
fields determined according to the Maxwell equation for "a cold" plasma 
must satisfy. 

The effective boundary conditions follow from the Maxwell equations, 
in which the currents are expressed by the fields by means of system (4) . 
For purposes of definition, let us investigate a wave in which the longi- 
tudinal field component is symmetrical with respect to the plane 
X = [E^,(,-x) = E^Cx), EL(-x) = -H , (x) ] (Ref. 2). As would be expected, 

the currents J^ and J^ — according to system (4) — retain the symmetry 

of the corresponding fields. Therefore, it is sufficient to obtain the 
boundary conditions for one of the plasma layer surfaces; they will be 
satisfied on the second surface due to the field symmetry. Let us find 
the boundary conditions which must be satisfied by the tangential field 
components on the x = 1 surface. Integrating equations containing J^ and 

J^, over an infinitely thin layer e(Z2. << e << 6, 1, where e — layer /86 

thickenss, 6 — depth of skin-layer) , we obtain 

i—t 

When the integrals in equations (5) are computed, it is advantageous first 
to perform integration over n : 



Xexpf-f (l-e-V)ldV- 
I— I L J l i_£ 
L J 1— • 



X exp 



77 



X exp {— $2 



"o^ ' 



){j£.(V)expr-'|(l+V)ld7,'- 

- Jf.hOexpFf (l+V)ldV|; 

; = .jfexp[-^^-fjJ^.(,Oexp^-fx 
V)j rfV - /P^ J f exp 1^ - S^ + 2^1 J £, (Y) X 

K exp P| (1 - e - vj') jdT]' + tp j' f exp (- ^2) { j £, (71 ') X 

L Jo _i 

I 

xexpf-f (1+V)W- |£^h')exp pi {l+V)]rfV}. 



X (1-e- 



(6) /87 



where only terms making the largest contribution in the case of Vr^ -> 
are retained. Performing integration over n' in equations (6), we find 

2iR 2iR 

asr, 



R 



2iR 2 



7«- + 



e ^ — p^e ^ 






r _£^ _lR-ri 

■ +|^£^(l)exp(-^^)^^d$32« 2^; 



e , ^ — p^e ^ 



2i;? 2i"R 



i n\ » 



»0 



2 /: = ^'Iexp(-$^)W^1Ip i£« 



2 



_ J J £^ (1) exp (- ^2) \d% -37W i« 



e ^ — p^e ^ 



(7) 



Since J^^ and J^ in equations (7) are proportional to Vrj,, when the integrals 
are computed with respect to C = — ^ in these equations, it is sufficient 

to confine oneself to the zero-th approximation with respect to v (i.e., to 



78 



strive to zero in the vj intervals) . Taking the fact into account that 
Imo) > 0, we finally obtain 



u>l(l~p)Ei{\); 



N»i 






(8) 



Thus, the effective boundary conditions taking into account the 
thermal motion of plasma electrons have the following form 






2 ]/^o>2/? 



««o(l+P) 






, W-. « .. KlUif. II -]- I/I IO)« II -1- Wl _,, I 



(9) 



Substituting the solution of the Maxwell equations for a plasma 
( I n I < 1) in a vacuum ( | n | > 1) in these boundary conditions 



ik^a 



where 



E'^ (71) --=Ach-Uri; E'J {-q) = -J A shxlri; H^ (vj) = -'-^A sh^l-q, 

2 

t: — «3 — e3« , S3 = 1 — -2 , « = — , 



(10) 



/88 



we obtain the dispersion equation 

S3 [X + .;. th X/] = ^; [(I + P) ^1 th X/ - i|l (, _ p J 



(11) 



In the case of 1 ->■ °°, this equation changes into an equation for half-space 
(Ref. 1). 

Thus, in the case of a plasma layer having a finite thickness, the 
surface wave damping is proportional to v.^. However, the wave damping in- 
creases with a decrease in the layer thickness , because the wave phase 
velocity decreases : 



AA3 = 



KiK„s|/3„3 



(1+P)kl,l'+^-^ 



; hal»l, e3<0; 



(12) 






79 



REFERENCES 

1. Gorbatenko, M. F. , Kurilko, V. I. Zhumal Teareticheskoy Fiziki, 

34, 6, 1964. 

2. Vajnishte3m, L. A. Electromagnetic Waves. "Soviet Radio" (Elektro- 

magnitnyye volny. "Sovetskoye radio"). Moscow, 1955. 



SINGULARITIES OF AN ELECTROMAGNETIC FIELD IN A /89 

NONUNIFORM, MAGNETOACTIVE PLASMA 

V. V. Dolgopolov 

As is well known, in an isotropic, nonuniform transmittant medium there 
is a sharp increase in the electric field strength of an electromagnetic 
wave at the point where the dielectric constant vanishes. This increase is 
caused by plasma resonance. This phenomenon, which has been called "infla- 
tion" of the field has been studied in several articles. It is pointed 
out in (Ref . 1) that the "inflation" must occur in a magnetoactive medium 
at the point where the refractive index for the wave becomes infinite. The 
phenomenon of field "inflation" for normal wave incidence on the layer was 
studied in (Ref. 2) for the case when a constant magnetic field was parallel 
to the plasma boundary. 

This article investigates the behavior of the electric field of an 
"inflated" wave close to the "inflation" point in a magnetoactive plasma 
for the case of an arbitrary angle between the direction of the constant 
magnetic field and the normal to the plasma layer surface, and for the case 
of oblique wave incidence on the layer. A solution is found over a wider 
region for certain special cases for a wavelength which is a little less 
than the distance at which the layer parameters change significantly. 

We shall assume that the parameters of the plasma magnetoactive layer 
change along the x axis, and the vector of the constant magnetic field lies 
in the xOz plane, forming the angle S with the z axis. The tensor of the 
dielectric constant e^j^ can then be written in the following form 

''ejcos^ + sgsin^ 9 kzCosQ (63 — ei)cos9sin ©N 

— /eg cos 9 , Ej /e2 sin ' ^^ 

(£3 — e^) cos sin — /sg sin Sj sin^0 + S3 cos^ 



80 



e/ft = 



The quantities e^, e2» £3 will be real, if dissipative processes are not 
taken into account. 

Assuming that the dependence of the electromagnetic field strength 
on time, y and z, is determined by the factor ei(-(jdt + k^y + k^^^ » ^^ 
obtain the following system of equations from the Maxwell equation for 
the electric field components 

(k^ - i ^11) E, + i (ky ^-^ + k^^-^) - ^ (e,,Ey + e.gE,) = 

— (kyk^ + '^^2^E^ = 0\, (2) 



iku 



/yfe^--e E 



dx 






{kykz + ^ HijEy + 



where k^ = k^ + k^. 
y z 



/90 



The following inequality is fulfilled in the vicinity of the "inflation" 
point for an "inflated" field 



dE 
dx 



>k 



This enables us to represent the field E in this region in the form of a 
series 

E = £(0) + £(') + 



where 



EO)\-^\kx\\E('>^^\E<'>^\ 



We shall always select the origin along the x axis, so that x = at the 
inflation point. We obtain the following from the two lower equations 
of system (2) 

£(0) _ £(0) _ 0. -^ = ikyEW + Cy.^=^ ik.EW + C; 



- I = ii^ f_ 

dx'' ^ dx c 



dx ~ ""^"^^ ' ""' dx 






where C and C^ are the integration constants, 



81 



Substituting the expressions obtained in the first equation (which 
is differentiated with respect to x) of system (2) (we shall disregard 

components containing ^ and ^^ ) and taking into account (1) , we /91 

8x 9x 
can write the equation for e( ): 

X 

I (HxE^^) + i2k,H3Eio, ^ const = 0. 

It can be seen from this equation that "inflation" can only occur where 
e^l vanishes. In the given approximation, it must be assumed that £^3 

is constant, and is limited to the linear component in Taylor expansion 
for Ell. Then the equation for eC^) assvraies the following form 



where 



11 dx \ x=o ' 



We can write the solution of equation (4) in the form of a sum of 
particular solution and the general solution of the corresponding homo- 
geneous equation. The particular solution is a constant quantity, and Is of 
no interest. The general solution of the homogeneous equation for the 
field E close to the "inflation" point yields the following expression 

c- ^ „— :i In (kx) (5) 

where A is a quantity which is constant with respect to x; 

0=2^. 



If a f= 0, we have 



^y 



where C^ and C2 are integration constants. 
In the case of a = 0, we have 

Ey = ikyA In {kx); Ez = ik^A In (kx). 



82 



Thus, if kg. = or e^ 3 = 0, the nature of the field "inflation" E 
is the same as in an isotropic medium. 

If k2 T^ and e-^^ i= 0, the closer the "inflation" point, the more 
intensely does the field E oscillate on both sides of it. Thus, the 
amplitude of the field component E increases as — , and the amplitudes /92 
of the components E and E remain finite. It can be seen from equation 

(3) that if ^^ = 0, at the point where en = "inflation" does not 

3 X 
occur in the case of k e , f= 0. 

Let us investigate the case when the medium parameters change slowly 
and the method of geometric optics may be employed when solving the system 
of equations (2) outside of the vicinity of the "inflation" point 

X 

E =A (x) e ° ' ^ 

Substituting expression (6) in the system (2) , we obtain the following 
equation of the fourth power for x(x) 

eiix" + 2fe,si3x3 + a W x^ + P (X) X + Y W = 0. (7) 

We shall not give the expressions for the coefficients a, 3, y, due to 
their ciombersome nature. 

If k^. and e-^g are different from zero, we have the following from 

equation (7) for the wave vector component x(x) of an "inflated" wave 
close to the "Inflation" point 

y~ - 2k^ "^' . 

This result coincides with formula (5) . The wave oscillates in space on 
both sides of the "inflation" point. 

1. If the magnetic field is parallel to the plasma layer surface 
(5=0), e,_ and B(x) vanish and equation (7) becomes a biquadratic equa- 
tion. We than have 

^2 _ k2 ii+i3 A.5 _ f! . /' -.^^l!i±!?^ + 

(8) 



±i;^[i^^-^^)[K-ii^ + i^l]' + ^'iK4^^- 



83 



Taking into account the components of subsequent order in the expansion 

with respect to the small parameter-r — /x2 in the system of equations (2), 

dx ^ 

we obtain the expression for the vector components A(x) : 



* z r2. ■ 



K] ' 



(9) 



where C is the integration constant corresponding to one of the four /93 
functions x (x) ; 

X^ky (S+C+ + r C-) + k. (C+ + C-) -/) - 2x ($+$- + ^^); 
^ = ^' [>^l - ? ^i) -c^ ^r- K^ = x^ + A^ - ^e,; 



If we may disregard dissipation, in the region where x is real, Y and Z 
are real, and the exponential factors in equations (9) characterize only 
the wave phase. 

As may be seen from (8), when e^ strives to zero (in the case of 



S=Oeii=ei) one value of x^ 
comes infinite as J^: 



goes to a finite limit and the other be- 



1 



£, = 



■k^ + H-2 



7^-2 



4 + K 



'i^[^i^^+U\ 



t,— 



In the case of x2 = ^-^i, the solutions of (9) are valid for the point 
where ej = 0. They are not "inflated". The remaining two solutions of 
system (2) for the vicinity of the point where ex = cannot be represented 
in the form of formula (9). When trying to determine them^ one may assume 
that £2 and E3 are constant, and for ei one may restrict oneself to the 
linear components in the Taylor expansion: 



t^x. 



After several ctmibersome computations, we obtained the following 
equation from the system of equations (2) for the field component E 



84 



Disregarding the last two components, we may write 

d?u 1^ 4 du 1 „ 

W^T-^-Tx — Y^^"' (10) 

where /94 

«=-a^: p = ~— -. . 

If we substitute u(x) = C~^v(5), x = ?^, equation (10) becomes a Bessel 
equation 

and, employing the recurrent formulas for the Bessel function, we obtain 

where A and B are the integration constants; Ii (2/ px) and K^ (2v'px) — 
the McDonald and Bessel functions. In the region under consideration, 
the components of the field E are related by the following relationships 

which enable us to determine E and E^, : 

E^ E 



^y=k;-~\Au{2Vrx)-BKo{2VVx)]. ^^2^ 



Taking the asymptotic form of the Bessel function and requiring that ex- 
pressions (11) and (12) change into expression (6) for large values of |x| , 
we may find the relationship between the coefficients A and B and the co- 
efficients C included in (9). In the case of p > 0, the coefficients C 
are related to A and B in the following way to the right of the "inflation" 
point (x is imaginary in this case): 



85 



-+• 



{- 






1 /"'^iV'i^— JtS 



where C = C^ in the case of x imaginary negative; C = C_ in the case of 
X imaginary positive. 

2. If the magnetic field is perpendicular to the layer surface /95 

IT 

(S = — ) , the system of equations (2) may be reduced to a system of two 
second order equations 









■i 



(13) 



where 



^. = p (ky^ + k^f): £z = i {k^ - kyf); E. = -^-i . ^ 



-A' 



dx' 



Outside of the vicinity of the "inflation" point, and representing (j) 
and ijj similarly to (6) in the following form, we have: 

X 

<5 -ndx 

?,'!' = A,, ^{x)e ° , 

We obtain the following from the system of equations (13) in the approxi- 
mation of geometric optics 



where 



Q = 



0)2 
r2 ^3 ■ 



■k' 



[(^^+^4)^3-^ft'=4] 



K' 



= .^+A^-^e,; 



86 



C is an integration constant corresponding to one of the four 
ftmctions x(x). 



As may be seen from (14), close to the point £3=0 (in the case of 

T . 

the following form 



= -rEii = £3) one solution for x'^ is finite, and the other solution has 



In the first case, solutions of (14) are valid at the point £3=0. 

They correspond to a "non- inflated" wave. In the solution of system (13), /96 
we shall assume that ei and £2 are constant, and we shall expand £3 in 
Taylor series in the vicinity of the point £3=0, and shall confine our- 
selves to the linear component 

We may find the following from the first equation of system (13) 

(15) 






Substituting this expression in the second equation of system (13) , 
disregarding the small components, and making the substitution v = 



-7~T > we obtain 
dx2 



d^v , \ dv . V „ 



dx^ '^ a: dx ^ "^ X 
where 

3 



When 5 = 2/ px is used as the Independent variable, this equation 
is reduced to a Bessel equation 






a-o , I av , f. 



Utilizing (15) , we find the following from the latter equation 

if = Ci//^" (21/^) + CM^ {2VTx), (16) 

where Cj, C2 are the integration constants; H^J^ (g) , h(§) (?) — Hankel 
functions . 



87 



As may be seen from (15), close to the "inflation" point (fi | 'v- 1 e 3ij; | << 
<< 1^1 — i.e., the contribution from (|) to the expression for the fields 
in this region is negligibly small. If we make the stipulation that (16) 
change into an expression like (14) (outside of the vicinity of the "in- 
flation" point) J we may find the relationship between the constants C2 , C2 
in (16) and the constants C in formula (14) . 

Just as in the case of an isotropic layer, the presence of "inflation" 
points in a magnetoactive , nonuniform, transmlttant medium leads to the 
fact that the thermal radiation intensity of this medium is on the order /97 
of the radiation intensity of an absolute black body, if the wavelength 
being studied is comparable to the distance from the "inflation" point 
to the layer boundary. When an electromagnetic wave, whose length is 
comparable to the plasma layer thickness, falls on this layer, the amount 
of energy absorbed by the plasma in the vicinity of the "inflation" point 
per unit of time (per 1 cm of layer surface) is 



^p2 (17) 

where E is the electric field amplitude in a vacuum. 



4jt 



However, (17) is valid until the phenomena caused by the oscillation 
nonlinearity and spatial dispersion in the vicinity of the "inflation" 
point exceed the phenomena caused by particle collisions in this 
region, i.e.. 



l(^V)y|:^|vy|; kY 



where v is the electron velocity caused by the wave field; v — the effec- 
tive frequency of electron collisions with ions; K '\- — '— — effective 

wave vector in the vicinity of the "inflation point; SL — layer thickness; 
m — electron mass; T - temperature. This leads to a limitation for the 
field E and the temperature T, which can be fulfilled by (17) : 

where Hg — the constant magnetic field strength; L — Coulomb logarithm; 
nQ — electron density in the plasma; e — electron charge. The maximum 
energy obtained by a charged particle per unit of time is 

W TTT-J • 

lOm'c/Io 



88 



In the case of hq = lO^^ cm-3, T = 10^ "K, Hq = lO^G, co 'v 5-10 
sec"-"^, w 'V' 100 ev/sec, E ;^ 1 v/cm. 



10 



Thus, it is not advantageous to employ "inflated" fields for plasma 
heating . 



REFERENCES /98 

1. Ginzburg, V. L. Propagation of Electromagnetic Waves in a Plasma 

(Rasprostraneniye elektromagnitnykh voln v plazme) . Fizmatgiz, 
Moscow, 1960. 

2. Denisov, N. G. Radiotekhnika i Elektronika, 1, 732, 1956. 



EXCITATION OF A MAGNETOHYDRODYNAMIC WAVE 
GUIDE IN A COAXIAL LINE 

S. S. Kalmykova, V. I. Kurilko 

As is known (Ref. 1), magnetohydrodynamic waves have low frequencies 

(to << fixT ^ 1.5'10^ , with respect to A,,^^ >> 10^ cm). At these fre- 

n v/ sec vac 

quencies, a coaxial line represents the most reasonable method for trans- 
mitting energy from an oscillator to a wave guide. Therefore, the problem 
of excitation of an axially symmetrical E-wave in a magnetohydrodjmamic 
wave guide by a TEM wave of a coaxial line is of great interest. If the 
plasma temperature may be disregarded, and its conductivity is great enough, 
under these conditions, the plasma may be characterized by the dielectric 
constant tensor without spatial dispersion. The problem under considera- 
tion is then a special case of the problem regarding the matching between 
an anisotropic dielectric wave guide and a coaxial line. Equations for 
determining the stray field were obtained in (Ref. 2, 3): 



'? ^'^ + 2 (nif j- [z in + z„ (/')] (f - J L^o ('") 2o C) J '' - 1' 



(1) 



~w^ j" 



+-" z,{nz(t' )dt' . inj^x) 

[z (/') + z„ (/')) {i'^ - k') {(' - • ^ 



89 



The diffracted magnetic field in the space between the wave guide 
(radius a) and the casing (radius b) is expressed by means of the in- 
determinate function (j>(t) E (})+(t) - (|)~(t) = <|)+(t) - <i>'^(-t) (the indices /99 
(+) designate the analyticity in the upper or the lower half plane of 
the complex variable t) : 

+~ 

//(z)== Y- • ^r-! • —I ?— r-(''+(0-4-9+(0-5^ X 

X ^"zr;^) exp (itz) dt for- 2 > 0, 



(3) 



where 



x+(/)_x+(-/) = -^<p(0 + 



7 if\^^ ^0 (a) . 7 //■> _ P (0 MM . 

A„(r)=/„(yr)^o(y6)-(-l)''/Cx(yr)/o(t;6), «=0,I; 



= (/2 _ /fe2)'A; p = 



";J-(^^-^^sJJ^ 



If the plasma density is large (uga^ » c^), equation (1) can be 
solved according to the iteration method. 

The reflection coefficient of a coaxial wave in the first approxima- 

tion, as a function of the parameter — - — - ■■ , , is 

a In(b/a) 






(4) 



When the amplitude of three-dimensional waves excited in a magnetohydrodynamic 
wave guide is computed, a distinction must be drawn between the four regions 
of the plasma parameters and the wave parameters . The magnetic field 
amplitudes have the following values corresponding to these regions (the 
amplitude of the coaxial wave equals unity) : 



'^'^=Eh)^^n + ^^-Hbla)\-\ 3,-l«-^«l; 



90 



II I I I II I 



-T'«e^-l«l: 



2 r >.y J 

^"=5n!^N- -?-«•«..; (5) 

Thus, the effective excitation of magnetohydrodynamlc waves by a 
coaxial line is only observed for strong magnetic fields (ej_ - 1 << 1) . 

2 
Harmonics with large numbers corresponding to the limiting case X >> 

>> a^/6^ are always only slightly excited. For weak magnetic fields, the 
main portion of the coaxial wave power is dispersed into excitation of 
the surface plasma wave. 

REFERENCES 

1. Glnzburg, V. A. Propagation of Electromagnetic Waves in a Plasma 

(Rasprostraneniye elektromagnitnykh voln v plazme) . Fizmatgiz, 
Moscow, 1960. 

2. Kalmykova, S. S. Uspekhi Fizicheskikh Nauk, 9, 2, 1964. 

3. Kalmykova, S. S., Kurilko, V. I. In the Book: Plasma Physics and 

Problems of Controlled Thermonuclear Synthesis, 4 (Fizika plazmy i 
problemy upravlyayemogo termoyadernogo sinteza, 4). Izdatel'stvo 
AN USSR, Kiev, 68, 1964. 



/lOO 



THEORY OF MAGNETOHYDRODYNAMIC WAVE SCATTERING 
AT THE END OF A WAVE GUIDE 

V. I. Kurilko 

The study of magnetohydrodynamlc waves is of great interest in 
solving several problems of plasma physics, such as controlled thermo- 
nuclear synthesis, magnetohydrodynamlc oscillators, etc. (Ref. 1). A 
great many articles have been recently published which investigated the 



91 



Illlllilll 



propagation of magnetohydrodynamic waves in unifoirm, unconfined magneto- 
hydrodynamic wave guides. However, practically every wave guide is con- 
fined. Therefore, it becomes necessary to investigate the phenomena re- /lOl 
lated to the scattering of magnetohydrodynamic waves at the end of a 
wave guide (for example, reflection of one of the eigen waves of such a 
wave guide, and its transformation into other waves). In addition, the 
theory of electromagnetic wave scattering includes the excitation of a 
confined wave guide. 

This article investigates these phenomena for a semi-infinite 
plasma wave guide. In the general case, this problem may be reduced to 
a system of two coupled integral equations or (in the presence of an 
infinite casing) to two infinite, coupled systems of algebraic equations, 
whose solution may only be found by numerical methods. Therefore, let 
us investigate the case when the end of the wave guide is covered by a 
conducting diaphragm. As will be shown below, the problem of determining 
the Fourier component of *]\e scattered field may be reduced to an integral 
Fredholm equation of the second type, whose solution may be obtained by 
a numerical method, and when there is a small parameter — in analytical 
form. 

We shall assume that the plasma conductivity is infinite, and its 
temperature equals zero. In this case, the electrodynamic properties 
of the plasma wave guide, as is known, may be characterized by the di- 

•, . "n ■, "0/2 Aire^n eHn 
electric constants ej_= 1+ ^, e „ = 1 - _ (^a,J = —^, '^ = ^, 

JL, = >> 0) I , which only depend on frequency. Since the frequency is 

^ Mc / 

fixed in our problem, the plasma wave guide may be regarded as a special 
case of an anisotropic dielectric wave guide. Therefore, let us first 
investigate the more general problem of electromagnetic wave scattering 
by the jump in the dielectric constants of an anisotropic dielectric wave 
guide, whose uniform sections are separated by the conducting diaphragm. 
The plasma wave guide parameters may be employed to find the analytical 
solutions of the general equation obtained. 

Thus , let us investigate an anisotropic dielectric wave guide 
(r < a, -<=° < z < +») with a piecewise-uniform tensor dielectric constant 

e(z>0. 0<r<a) = {eO), eO)); 
s(z<0. 0<r<a) = (e(2). e<?)); 

A 

e(— oo<z< + co, a<r <b)^\. 



92 



Let us assume that the uniform sections of the wave guide are /102 
separated by an ideally conducting diaphragm (z = 0, < r < a), and 
that the conducting casing is not confined and is located at the distance 
r = b from the wave guide axis. Let us assimie that one of the eigen, 
axially symmetrical E-waves of this wave guide, which is characterized 
by the wave number hjjj, falls on the nonuniform section from the right 
wave guide. Let us determine the amplitudes for the eigen waves of 
both wave guides which are excited due to scattering of this wave. We 
may write the solution for the fields in the following form 



+ " 



//« = ^ //(OAi {t)e"'dt + exp{~ih„z) 



\ 



£| = j" f //x (0 A„ (0 e"^ dt + Zl exp {-ihrnz) 



— oo <2< 

< + oo; 
t^n(t) = ln(va)KAvb)-{-l)''Kn{va)U{j:b), n = 0.1: 



r = a + 0, 



(1) 



u (0 = (/2 — A2)v.; k = '^, lmo>>0; 



Hl = \K (0 h (PsQ) e^'dt + 2A, cos A„z 

E\ = t > (0 hs (0 Ji (Psa) e^^dt + 2A,Zl cos hmZ 



?; =. J ^ /j. (0 Ji (Psa) e"' + 2Ash„ (sin h^z) ^ 

CO ^ ) 



r = a — 0, , 
s = 1 — z> 0, 
s = 2 — 2 < 0, 



(2) 



where 






Ai = 1; Ai =0. 



The fields thus selected satisfy the Maxwell equation. We can deter- 
mine the remaining, unknown Fourier amplitudes H, hg of the desired fields 

from the boundary conditions on the lateral surfaces of the wave guide /103 
and on the conducting diaphragm. The boundary conditions on the wave 
guide — vacuum surface have the following form 



93 



lilllli 



//y' = '^%' ^i'* = £^ f or r = a, 2 > 0; 

Hf = //J. £f = £', for. r = c. 2 < 0. (3) 

Substituting (1) and (2) in the boundary conditions (3) and employing 
the results derived in (Ref . 2) , we may express the vinknown f vinctions 
H and hg by the boundary values on the contour Im t = of the functions 

which are analytical in the upper (+) and the lower (-) half-planes of 
the complex variable t : 

K it) A (ha) = ^) [z it) T+ it) - 1+ (0 - 2^.7(7-^!)} ' 
h. it) A (M = ^ {z (/) X- it) - r it) - 4r^} ' ("^^ 

// (/) Ai it) = ^1^^ [z^ it) <p+ (0 - ^+ it) - 2.V(!Im 1 = 

D,it)~Z^it)-Zit), Z(0=f • J^. 



where 



The latter equation (4) represents a boundary problem for determining 
the unknown functions (j)"*", \^'^, K~ and x~. The relationships lacking between 
these functions may be determined from the boundary conditions on the con- 
ducting diaphragm 

£^(2 = 0; 0<r<a) = 0. (5) 

Substituting E^ in the boundary conditions (5), we obtain 

h'i—t)^h'it). (6) 

We obtain the following by means of the latter equation and the Sommerfeld 
condition for the finiteness of the magnetic field and integrability of 
the electric field close to the diaphragm edge: 

^-it)~-<ffi-t); S7(0=-t^(-0. (7) 

where 



/104 



94 






Im hm > 0. 

(Equations (6) are only necessary for fulfilling the boundary conditions 
(5). We shall assume that these equations are sufficient, although this 
may only be proven for a rectangular wedge [Ref. 3]). 

Thus , the boundary problem for determining the function assumes the 
following form 






(8) 



This type of boundary problem for one special case (ei = 1, £2 ^ ") 
was first studied in (Ref. 4). It was shown that in the presence of a 
casing, when the coefficients have singularities of only the pole type, 
it can be reduced to an infinite system of algebraic equations. However, 
by employing the formulas of Sokhotskiy — Plemel' , it is more advantageous 
to reduce the problem (8) to an integral equation. By combining and sub- 
tracting equation (8) with its mirror image at the point t = 0, we obtain 
(Ref. 5) 

1 (hsn_^ , _L z / A V ^(\ JL_ _ 1 ]h (n ^.. _ 

2 (9) 

A„ Z (t) \y Z' (t) — Z' 

--if -,-13^2, o[^t) ^ /W= 2(0ti(0 = '1'x(0. 

5=1 

The index of the latter equation equals zero (Ref. 5). Therefore, it is /105 
is equivalent to one integral Fredholm equation of the second type for 
'J^l(t) = .j^l(t) - i^+(-t): 



95 



+ - 2 + = 



(10) 



S=I 



ij^,^/r) 



According to the general theory of singular equations with a kernel 
of the Cauchy type (Ref. 5), the solution of (9) which vanishes in the 
case of t ->• " exists and is real. Due to the equivalence between (9) and 
(10) , the same holds for the solution of equation (10) . It can be solved 
numerically in the general case, and in the presence of parameters — it 
can be solved analytically, even if there is no casing (b ^ «>) and the 
coefficients in equation (8) have singularities of the point branching 
type. 

In the case of the magnetohydrodynamic wave guide, which we are in- 
vestigating, the ratio between the wave guide radius and the wave length 
in a vacuum serves as one of the large parameters. Even for Hg '^ 10^ gauss, 

Xq ^ 7^ '^ 10^ cm. Therefore, in the case of a ^ 5 cm the ratio a/X (A — 

wavelength in a vacuvmi) cannot exceed 10"^ (with allowance for the require- 
ment that X >> Xq), so that even In (X/a) is large (In (X/a) ^4). In 

addition, in a significant number of important cases the linear plasma 
density na^ is great, so that the ratio — = is small. 

Assxmiing that the inequality 6 << a << X is fulfilled, we can signifi- 
cantly simplify equation (10) . Disregarding terms on the order of /106 

(6/a)In(X/a) , (— ) In(X/a) and taking the fact into account that £2 - 1» 



•(f) 

my t'-t '. Ki yzimt'-t)- r:i ' t'^_hi[+ z(t)\ 



we obtain , , 



(10') 



(the index «: 1 » for i|) is omitted from this point on). Equation (10'), as 
may be readily seen, is equivalent to the scalar boundary problem 

<j.+ (0=z(/){"-(/)-^-^[i + ^]}. ^^^^ 

The solution of (11) has the following form 



96 






where 



^■'(0 = f7S^-(0; ^+(0 = 3p^. 



and the singularity at the point t' = t passes around the contour C from 
below.. 



Within an accuracy of a term on the order of { — \ In — << 1, we 
have \x; a ' 



(?) 

X+ (t) = (to + kayf' exp fc C y=^. ' FT7V X 

I ka^ (12') 

xln[ln(|-nn|)]|. 

where y is the Euler constant, and in the case of In — >> 1 we have 

a 

^"(') = [|-nn|]"'«» + .«)v.{.-|.!i«±f!). 

1 xl"i 

Expansion in powers of In — is usually employed in antenna 

I ^ I 

theory (Ref. 6), and numerical integration may be employed to determine 
expression (12') for YX+(t) which are not too small. 

In the same approximation (6 << a << A) , the boundary problem for de- 
termining the function <j)+(t) has the following form 

T+(0=,4^--(0-S'-^[l+^]. (13) 

The integral in the solution of (12) may be readily computed by closing 
the integration contour in the lower halfplane in the first term, and 
in the upper halfplane in the second term, by the complex variable /107 
t. The integral in the solution of (13) may be computed similarly for 
(()+(t) . Substituting the expressions thus obtained for i('''"(t) and <|i"'"(t) into 



97 



J 
i 



the formula for the amplitude of the Fourier-field in a vacuum 

Ax (0 ^(0 = 5;^ {2^ (0 ^+ (0 -1'+ (0 + ^ • ^-7^]' (1^) 

by means of (1) we may find the expression for the amplitude for a wave 
with the number 1, which is excited in a wave guide (z > 0) during the 
scattering of a wave with the number m: 



R,--^^'fi'-^^+^- 






^ + 



(15) 



^/2;L 1 „ _km 



ni,m^=^- 



whereDu = ^D,|,= ft,;Z;„ = Z^(A,J;Xt„ = X+(A,„); 

The latter expression may be significantly simplified in the most in- 
teresting cases of small and large retardations njj^ ^. It thus appears that 

the sum in the parentheses. in (15) is represented in the form of a power 
series of the logarithms for the ratios X/a and X/6. Retaining the old 
terms in this series, we obtain 



^>Vr^''"^- (16) 

^1 (v/. m) = 0; ni = Sj.; 
r »S ]' In (e, - 1) ^ 

°('x-') 



98 



^«nx-l«l; (17) /lOS. 



«*vf "j." a an'/' 



In - In* 

a an^ 

-^«l«nx«^; 

Xa . 1 . X . X 



(18) 



In -jj In - -I- In ^- In 






52 <$.>/, m <5. p- (19) 



It can be seen from these formulas that the coefficients of the inter-trans- 
formation of the magnetohydrodynamic waves Ri decrease with an increase 

in the wave number and a decrease in the magnetic field strength. If the 
plasma wave guide is surrounded by a conducting casing r = b, -«> < z < -H", 
it is impossible to study magnetohydrodynamic waves in the case of X >> b, 
and only the transformation of one oscillation into another occurs at the 
end of the wave guide. The transformation coefficients for this case may 
determined by substituting X+ (t) from the solution of the corresponding 

problem X + (t) = valn^ X - (t) and the wave numbers h^^ui — from the 

solution of the dispersion equation, in expression(15) . 

D, {h,y. = Z\- ^' , > In I = 0.. 



REFERENCES 

1. Artsimovich, L. A. Controlled Thermonuclear Reactions (Upravlyayemyye 

termoyadernyye reaktsii) . Fizmatgiz, Moscow, 1963. 

2. Rapoport, I. M. Doklady Akademii Nauk SSSR, 59, 1403, 1948. 

3. Kalmykova, S. S., Kurilko, V. I. Doklady Akademii Nauk SSSR, 154, 6, /109 

1964. 

4. Jones, D. S. Proc. Roy. Soc, A215, 153, 1953; Philos. Trans. Roy. 

Soc, A247, 499, 1955. 



99 



5. Noble, B. The Wiener-Hopf Method. Izdatel'stvo Inostrannoy Litera- 

tury (IL), Moscow, 1962. 

6. Muskhelishvili , N. I. Singular Integral Equations (Singulyarnyye 

integral ' nj^e uravneniya) . Fizmatgiz, Moscow, 1962. 

7. Vaynshteyn, L. A. Zhumal Teoreticheskoy Fizikl, 29, 673, 689, 1959. 



DETERMINATION OF PLASMA TEMPERATURE AND DENSITY 
DISTRIBUTION BY REFRACTION AND DAMPING OF A BEAM 

V. L. Sizonenko, K. N. Stepanov 

Let us Investigate the passage of a beam through a nonuniform plasma. 
We shall show that, by changing the angle of incidence or the frequency of 
the microwave signal, we can determine the plasma density distribution 
by the beam refraction, and can determine the plasma electron temperature 
distribution by the damping of the wave energy along the beam. 

In order to determine the beam trajectory in a plasma, let us employ 
the Fermi principle 

8pds=0, (1) 

a 

where k is the wave vector; ds — an element of length along the beam tra- 
jectory. In an isotropic plasma, we have 

k^^y^^-wl, (2) 

where o) is the wave frequency; ojo(i^) * W ^"^ — plasma Langmuir fre- 

f m 

quency; n(r) — plasma electron density. Modulation is performed in 

formula (1) for fixed ends (^x^ a '^ ^^i b ~^' Taking the fact into account 

that ds = vg^^dx^dxi^, from expression (1) , with allowance for (2) , we may 
obtain the equation which determines the beam traj ectory for given wq (r) /IIQ 



100 



ds 



I 1/~1 2 ^^A^ 1 1/^ 2 ^Skl <^>'k ^^i 5 -1/-^- 2 „ 



By employing equation (3), let us investigate the inverse problem: 
Let us find wq (r) , i.e., the plasma density distribution based on the beam 
refraction, as a function of the angle of incidence for a fixed frequency, 
or as a function of the frequency for a fixed angle of incidence. 



Planar Problem 

If the density of a plasma filling the halfspace x > depends only 
on one coordinate x, and it increases monotonically with an increase in x, 
then the beam trajectory in the case under consideration is flat (Figure 1) 
Equation (3) may be written as follows 



^ ' (4) 



•= Ydx'' + dz'' 
Integrating equation (4) , we obtain 



]/ (1)2 — cBo^ = a = const, ^^^ 

where a = to sin )|;, ijj is the angle of incidence. 

The position of the point at which the beam leaves the plasma z = I 
can be determined from the plasma density distribution; it depends on the 
angle of incidence and frequency. If the dependence i,{^) is known from 
experiments, we may obtain the density distribution n(x) in the case of 
X < X*, where aiQ (x*) = w (this problem is similar to the problem of deter- 
mining the potential energy by the specific dependence of the oscillation 
period on energy [Ref. 1]). In actuality, it follows from equation (5) 

that 

dz sirn]^ 

dx 



/ 



<^os'*-^ (6) 



We thus have 




/111 



(7) 



101 



where x = xq is the rotation point of the beam, determined from the 
following condition 

cos'' tb = i— . 

* (1)* 

2 

Introducing a new variable u = ~ instead of x, from formula (7) we 

obtain cos"<i> 










It thus follows that 

I 7 cos'iji 



(8) 



-^4i^L = +2\dcos«a ^^'" 

sin-f-y^— cos2<j; J ^J K (cos2 4. — u) (f — cos'ii;/) 





7 

I 



2^j|rf« = + 2u^(T). 



2 

too 

Assuming that y = ~r» we finally find that 

0)^ 




-^ ^^-=. (9) 






COS^l}' 



Equation (9) determines the function x(a)o) » i.e., the dependence of density 

on the coordinate in implicit form, according to the specific dependence 

l{ii) . We may employ this equation in the case of a cylindrical plasma, 

when the plasma is probed by a beam in the plane ^ = const passing through /112 

the cylinder axis. Thus, oig = wgCr) and x = r = R, where R is the plasma 

radius (uqCR) =0). By performing similar measurements for different 

values of the aximuthal angle ()), we may obtain the density distribution 

in the case when density depends on the angle ^ (but does not depend on z) . 

The plasma density distribution may also be determined by probing the 
plasma with microwave signals having a different frequency. For a given 
angle of incidence, the position of the exit point of the beam z = £ de- 
pends on the frequency co. Knowing the function £((0) from experiment, we 

102 




Figure 1 



may readily determine the function oIq (x) according to equation (7) . It 
follows from equation (7) that 




2<s>dx 



\ COS^lJ'' 



= 'r:x{^). 



(10) 



0)0 
Assuming that y= 7> 

" COSlI) 

(joq (x) in implicit form 



we obtain the expression determining the dependence 



costj/ 



X = 



1 p / (a.) dai 



(11) 



We may also obtain formula (11) from the expression given in (Ref . 3) 

[see also (Ref. 4)] for the real layer height x((0o) according to the specific 

effective height Xg(w) for the case of normal wave incidence on the layer, 

with allowance for the following relationship (Ref. 4) 



l{'^)=2tg'^Xg{i>^cosj>). 



Cylindrical Problem 

If the beam passes through a plasma cylinder in a plane which is per- 
pendicular to the cylinder axis (Figure 2) , equation (3) may be represented 
in the following form 



i(/-=']A;;^:r75|) = o; ds=.Vd?^T^^' 



(12) 



103 




Figure 2 



The plasma density distribution n(r) may be found from the specific 
experimental dependence of the angle at which the beam leaves the plasma 
(jjQ on the angle of incidence i|/ * (this problem is similar to the problem 
of determining the potential energy from the specific dependence /113 
of the scattering cross section on the scattering angle (Ref. 1, 2). Inte- 
grating equation (12) , we obtain 






sin^ <j» , 



(13) 



It thus follows that 



cpo(<}')=2/?sin^ 




rfr 






(14) 



where r = rg is the rotation point of a beam which is determined from the 
following condition 



Assuming that 



* This problem was solved in (Ref. 5), 



104 



1 ^ (t "oU* 



we can write equation (14) In the following form 

stn'<|< 



,p.(« = 2slntj j7j=g;y| 



J y^u — sin* ^ 



(15) 



from which we find that 
1 



1 1 sin»<(i 



du 



du 



/(sin* i|/ — f) {u — sin* <}/) 






Assuming that « = y* = 1 1 §)^i we may find the dependence r = r(a)o) 



r = R exp 




sin^ (p — " 



(16) 



Formula (16) determines the function r = r(a)o). I.e., the distribution /114 
of density over the radius In implicit form, according to the specific 
dependence of the beam exit angle on the angle of Incidence. 

Performing measurements In different planes z = const, we may thus 
obtain the density distribution along the cylinder axis in the case of 
axial symmetry. For a fixed angle of incidence \l>, equation (15) determines 
the beam exit angle <^q as a function of frequency. A knowledge of the 
function <|)o('»)) enables us to determine the density distribution. We thus 
assume 



2(r) = 



,(r) 



E = arc sin I - sin ^j . 



(17) 



Equation (14) may then be written as 



105 



Illlllllllll II I 






2W JYu>-^ 



Qi- 



We thus find 



' -' ' 2a. JdQ 





Since 5(0) = ^io-n ~ 'l^» assvming that y = Q, -we can write 



(as) du> 




We thus obtain the expression determining the dependence n(r) in implicit 
form 



1^ p ?o H da> I (18) 



-f + i|iJ^.) 



Temperature Determination from Beam Damping 

Due to collisions , the energy of a beam leaving the plasma is 
eT times less than the incident energy, where t = / x ds is the 
optical plasma thickness; x — damping coefficient. If the wave fre- /115 
quency is considerably greater than the frequency of collisions of elec- 
trons with ions and neutral particles v (r) , we then have 

x(r) = 



•■l/--¥ 

Knowing x(r), we may find the plasma temperature distribution from the 
specific dependence of v on temperature. 

The damping t depends on the angle of incidence f. We may find the de- 
pendence of the damping coefficient on the radius from the quantity t(ijj), 
which is measured experimentally, and from the specific density distribution. 

Let us first investigate the planar problem (in the case of a cylin- 
drical plasma, the beam trajectories lie in the plane ((> = const). For 



106 



T (ip) , we may write the following expression 



dz 
Substituting -J— from formula (5) In this, we obtain 

x(.j))=2 1 x(a:)i/ "7"" , dx==2 \ X V '~" du. (19) 



Let us Introduce the function f(u) according to the equation 

(20) 



We then have 


xr<];)-2 ^" 




J ]/ 00524/ — u 




w = ^Ul/ i-;r»/\ i/io_ 



(21) 



Solving this equation and equation (8) , we find 

u 
2 

Assuming that A = 2» we obtain jWb 



2\ p _a^^il^'l_ (22) 



-2 — cos* 4' 

..2 ' 



In explicit form, this formula determines the dependence of the damping 
coefficient on coordinate x. 

If the beam trajectories lie in the z = const plane, we then have 

R 

t(^)=2j x(r)|/"l + r^(g)'dr. 



107 



Substituting the value of -^ from formula (13) , we obtain 



^(<1>)=2\ x(0 




„2_<«2_<„2^sin*4- 



dr. 



(23) 



Assuming that 



„ '■' (a "A df(u) / V r 1 /" , »o dr 



we may write equation (23) in the following form 

1 



J Yu — sin' 
sin' 4« 



(24) 



Solving it in the same way as equation (19) , we find 



r-^l—dsin'O/ 



Assuming that y = u(r), we obtain the following expression 



:(r) = 



dVu \ 
<^ J 



di 



dY u \ ds'\ni/ 



dsin<j/ 



dr J Ysm^if — u 






/117 



(25) 



Expression (25) determines the dependence of the damping coefficient 
coefficient on the radius according to the specific dependence of x on i|;. 



REFERENCES 

1. Landau, L. D. , Lif shits. Ye. M^ Mechanics (Mekhanika) . Fizmatgiz, 

Moscow, 1958. 

2. Firsov, 0. B. Zhumal Eksperimental'noy i Teoreticheskoy Fiziki, 24, 

279, 1953. 

3. Rydbeck, 0. Phil. Mag., 30, 282, 1940; 34, 130, 1943. 



108 



4. Ginzburg, V. L. Propagation of Electromagnetic Waves In a Plasma 

(Rasprostranenlye elektromagnltnykh voln v plazme) . Fizmatgiz, 
Moscow, 1960. 

5. Smoys, S. Joum. of Appl. Phys., 32, 689, 1961. 



109 



& 



SECTION III 
PLASMA NONLINEAR OSCILLATIONS AND WAVE INTERACTION /118 



EXCHANGE OF ENERGY BETWEEN HIGH FREQUENCY AND LOW FREQUENCY 

OSCILLATIONS IN A PLASMA 

V. D. Fedorchenko, V. I. Muratov, B. N. Rutkevlch 

Oscillations with frequencies v on the order of 100 kc (Ref. 1), 
as well as oscillations at the electron plasma frequency (a3oe ^ 50 Mc) , 
occur in a plasma produced by an electron bundle in a longitudinal magnetic 
field at a pressure of 10-^ — 10"^ mm Hg. It has been found that there 
is a relationship between the low frequency and high frequency oscillations 
(Ref. 1-3). High frequency oscillations cause an increase in the low 
frequency oscillation amplitude. On the other hand, they undergo ampli- 
tude modulation at a low frequency, which leads to the occurrence of com- 
bined frequencies w + v. The amplitudes of oscillations having the fre- 
quencies 0) + V and o) - v are not the same, which — just as with the 
Landsberg-Mandel ' shten-Raman effect — can indicate the direction and 
effectiveness of energy transfer from oscillations of one frequency to 
oscillations of another frequency. 

As has been shown previously (Ref. 1), low frequency oscillations /119 
represent ion motion (which is transverse with respect to the bundle) in 
the field of the space charge of electrons contained by a magnetic field. 
The frequency of these oscillations may be readily computed from the 
following simple model. Let us assume that electrons and ions fill one 
and the same cylindrical region in the equilibrium state. The simultaneous 
shift of all ions by the quantity r^, and of electrons by the quantity r^, 

leads to polarization, which can be determined as follows in the case of 
a small shift 

(1) 

where n is the plasma density. 

We can write the following expression for electrons and ions 



110 



I ■ III! ■ I 



or, with allowance for (1) , 



r; = — (Boi {n — re) + — [nB]; 
nil 



r, = o)o« (r/ — /-,) — — [r^B], 
"'t 



(4) 
(5) 



where 



ne- 



..2 



"' 'o""' (6) 



0« 



i2 



e„m, 



o'"t (7) 



Disregarding the electron shift during orbital motion, we shall only 
allow for their drift In crossed electric and magnetic fields , and 
we shall designate the position of the electron guiding center by the 
vector Tg. According to expression (5), we then have 



e r r., e 



ri — r,=-- J- ['■'^1 = r f''«'^l- (8) 



/n,-<o, 



i"'Oi 



It follows from formulas (4) and (8) /120 



e 



^'=-^f^'^J+i-.f^5I.' 



and thus 



^' = - ^ t^'^J + h. l^^l = -^ a^-S] B\ = 



e>B' 



i^oi ' '«,«>o/ ' (9) 



or 






(10) 



where 



2//=-^' (11) 



"o< 



— the ratio between the ion cyclotron frequency and the plasma frequency. 
Substituting the expression obtained for r^ in formula (4) , we obtain 

n-^CnB] + <o„^,(l + V)^ =0. (12) 

Equation (12) has a r-^ 'V' e type of solution, and the frequency 

111 



must satisfy the following equation 

Q^ + QhQ~1— Q-2 = 0. 



I.e. , 



2i2 = -|-^±l/ ^f + I+V 



(13) 



(14) 



In the case of f^g ^ 1, v ^ oiQi "^ •%/ "T* 




m 



X 



If ^TT >> 1, equation (13) has two roots; 



Qi = — Q„ (15) 



22 =J- 



Q^- (16) 



If % << 1 Q=±i'. 



(17) 



The oscillation frequency v decreases with an increase in the magnetic 

field strength, which coincides with the solutions of (16) and (17), if 

it is assumed that the frequency oJq. depends slightly on the magnetic field 

strength. If cOq. does not depend on B — i.e., Wq. is constant — it may /l^l 

be readily determined according to any pair of measured v and B, which 
would then enable us to compile a graph showing the dependence of v on B 
according to formula (14) . The dependence of ion oscillation frequency on 
the magnetic field strength is shown in Figure 1. The curves were re- 
corded for different current values in an electron bundle (1 - I = 10 ma, 
2 - I = 30 ma) . The dashed curves are drawn through the points computed 
according to foirmula (14) , under the assumption that the plasma density 
does not change when there is a change in the magnetic field strength. 
The computed points do not lie on the experimental curves (particularly 
in the region of small fields) , which is no doubt related to the variability 
of (J^oH • Nevertheless, it may be stated that the nature of the dependence 

is correctly imparted by our simple model. 

Modulation of high frequency oscillations may be due to low frequency 
oscillations of the plasma density. Let us assume, for example, that the 

plasma is located in an external field Ee-"-*^*-, which is transverse with 
respect to the bundle, and the frequency w is so great that only allowance 
for the electron component shift has any meaning. The external field 

ggiojt gives rise to a shift and the occurrence of a polarized field 

112 



'\< — 'r„ . The equation for electron motion in these fields may be written 
in the following form 



^~—ip-'+f-)-il'-S>- (18, 



We obtain the following expression for forced electron oscillations 



e pAoit 



--1-Ee 






+ »(">//« — "^ 



(19) 



If it is assxjmed that the plasma density undergoes oscillations at the fre- /122 

2 2 

quency v << w and the quantity Ugg has the form tOgg(l + a cos vt) , while 

a << 1, then expression (19) changes into the sum of the oscillations with 
the principal frequencies (oi) and the combined frequencies (to + v) : 

ete , a e t e 






"^ 2 ■ -/ eg. 



-; /((u—v) < 
e £ e 



(20) 



The oscillations have resonance close to the electron cyclotron frequency. 

This model is inadequate for determining oscillation intensity at the 
combined frequencies. 

The experiments were perfomaed on a hollow electron bundle in a longi- 
tudinal magnetic field with a strength ranging between 200 - 2000 oersted. 
The bundle length was 50cm; diameter — 2 cm; energy — 250 v; and the 
current — 20 - 40 ma. The bundle was located in a metallic tube having 
a diameter of 9 cm. Pressure in the chamber was several units of 10~° mm Hg. 
The interaction between the outer, high frequency field and the low fre- 
quency ion oscillations was studied. The interaction was observed in 
three cases: Vlhen the frequency of the outer signal (1) did not coincide 
with any of the plasma eigen frequencies, (2) coincided with the electron 
cyclotron frequency, and (3) coincided with the plasma electron frequency. 

In the nonresonance case, the outer field was transverse with respect 
to the bundle. It was produced between two conductors having a diameter 
of 0.2 cm and a length which was close to the bundle length. The conductors 
were parallel to the bundle axis at a distance of 6.5 cm from each other. 
Figure 2 shows the diagram of the experimental apparatus employed to /123 

113 




00400 aoeoo 0.1200 w."'" 



Figure 1 



study the interaction between ion oscillations and the outer electric 
field, whose frequency does not coincide with any of the plasma eigen fre- 
quencies (1 — solenoid; 2 — electron gun; 3 — probe; 4 — hollow elec- 
tron bundle; 5 — collector; 6 — conductors between which a high fre- 
quency field is produced; 7 — coils connecting the generator and the 
measurement circuit. The resonance curve for the circuit producing the 
high frequency field is shown in the upper right) . The capacitor is a 
section of the resonance circuit weakly connected to the GS-23 generator 
and the C4-8 spectrum analyzer. The resonance frequency of the circuit 
was 13.88 Mc. The resonance barely shifted when the bundle was switched 
on. The resonance width was quite large (Q = 50), so that the combined 
frequencies did not go beyond it. The connection between the circuit, the 
generator, and the measurement circuit was selected so that the resonance 
curve was symmetrical. This was important for comparing the intensities 
of combined oscillations with the frequencies cj + v and 03 - v. 



Figure 3 presents typical spectra of oscillations produced when the 
resonance frequency was used (the spectra were obtained under the following 
conditions: Pressure p = 3.4'10~^ mm Hg, electron bundle current I = 30 ma, 
anode voltage U =250 v, effective variable outer field strength v^ = 64 v, 
magnetic field strength H: a — 600 oersted; b — 780 oersted; c — 920 
oersted) . Lateral frequencies spaced at the frequency of ion oscillations 
may be seen, in addition to the frequency employed. Employing the termin- 
ology used in the theory of combined scattering, we shall call the lateral /124 
lines red (o) - v) and violet (u + v) "companions". The relative height 
of the companions (with respect to the carrier height) was 1-2%, and it 
increased with an increase in the amplitude of ion oscillations and the 
amplitude of the outer signal. 



The heights of the red and violet companions, generally speaking, 
were different, and this difference depended on the magnetic field 



114 



rt-M-e 




TW^ KC^-a 



Figure 2 



strength. As may be seen from Figure 3, when the magnetic field strength 
was 600 oersteds, the violet companion was higher, and for 920 oersteds 
the red companion was higher. At a field strength of 780 oersteds, the 
companions were the same. In order to explain the difference in the 
heights, we should point out that the increase in the ion oscillation 
amplitude when a high frequency outer signal was employed was always 
greater for large magnetic fields. For small magnetic fields and the 
same intensity of the outer signal, the amplitude increase in ion oscilla- 
tions disappeared. Thus, an increase in the magnetic field strength can 
improve the conditions for transferring energy from high frequency oscilla- 
tions to ion oscillations. 



Comparing this with the data given in Figure 3, we may arrive at the 
conclusion that there is a relationship between the companion heights and 
the direction of energy transfer. The predominance of the red companion 
corresponds to the transfer of energy from high frequency oscillations to 
low frequency oscillations; the predominance of the violet companion 
corresponds to the energy transfer in the opposite direction. The direc- 
tion of energy transfer no doubt depends on the relationship between the 
amplitudes of the ion oscillations and the outer signal. In actuality, 
an increase in the magnetic field strength decreases the ion oscillation 
amplitude, which leads to a more effective energy transfer from the high 
frequency to the low frequency for a given amplitude of the outer signal. 

The same result may be achieved in another way: by changing the 
outer signal amplitude for a constant magnetic field. With an increase 
in the outer signal, the red companion becomes higher as compared with 
the violet . 



/125 



Let us turn to an experiment in which the outer field frequency 
coincides with the electron cyclotron frequency. Figure 4 shows the 



115 




Figure 3 



diagram of the experimental apparatus for studying the interaction be- 
tween ion oscillations and the outer field at the electron cyclotron fre- 
quency (1 — solenoid; 2 — electron gun; 3 — probe; 4 — hollow electron 
bundle; 5 — collector; 6 — volumetric resonator; 7 — connection with 
the oscillation source and measurement circuit) . The outer field was 
produced in the volumetric resonator which encompassed almost all of the 
bundle. The mode Hj^, was excited in the resonator at a frequency of 

2265 Mc. When the bundle was switched on, the oscillation level in the 
resonator sharply decreased, when the magnetic field strength reached a 
value corresponding to electron cyclotron resonance (Figure 5) . The 
width of the resonance absorption curve was primarily determined by the 



116 




J— KS4-5 



/(gs-22 



Figure 4 



A.rel 
lunit 




QOBOO 0.1200 H,mJ!, 

Figure 5 



nonuniformlty of the magnetic field over the length of the system. Compan- 
ions appeared close to resonance which coincided with formula (20) . The 
violet companion was higher than the red companion, which was apparently 
related to the small amplitude of the signal supplied. 

Let us examine the case when the outer field frequency coincides 
with the electron plasma frequency. The outer signal is supplied to a 
grid located at the bundle origin. Measurements are performed by a probe. 
Figure 6 shows the spectra for several frequencies close to the electron 
plasma frequency (p = 4-10-^ mm Hg, I = 40 ma, U = 250 v, H = 1000 oersted, 
voltage on the grid u^ = 0.1 v. Oscillation frequencies of voltage on the 

grid: a — 39 Mc; b — 40 Mc; c — 41 Mc; d — 42 Mc; e — 43 Mc) . It 
can be seen that the interaction is resonant in nature. The heights of 
the companions are large (the total altitude of the carrier is shown in 
the photographs) , which points to the effectiveness of the interaction be- 
tween ion oscillations and the outer signal at the electron plasma fre- /126 
quency . 

The amplitude of the signal supplied is 0.1 v , which explains the 



* The effective voltages are always employed. 



117 







Figure 6 



predominance of the violet companion. When the amplitude of the outer 
high frequency field increases (for a magnetic field strength of 1000 
oersted), the red companion begins to predominate. A similar relationship 
between the companion heights and the amplitude of the outer field is ob- 
served for H = 7000 oersted. With an increase in the red companion, the 
altitude of the principal line (carrier) decreases. If H = 700 oersted, 
a decrease in even the principal line may be observed when the outer 
signal is intensified (from u^. = 0.5 v to u^ = 0.95 v) . For a compara- 
tively small magnetic field (H = 400 oersted) and a significant outer 
signal amplitude, there is very strong interaction which is accompanied 
by the appearance of many combined frequencies . 

These data point to the effective transfer of energy from electron 



118 



plasma oscillations to Ion oscillations. This phenomenon may probably 
be employed to increase the energy of the plasma ion component. 



REFERENCES 

1. Fedorchenko , V. D. , Rutkevich, B. N. , Muratov, V. 1., Chernyy, B. M. 

Zhurnal Teoreticheskoy Flziki, 32, 958, 1962. 

2. Fedorchenko, V. D. , Muratov, V. I., Rutkevich, B. N. Zhurnal Teoreti- 

cheskoy Fizikl, 34, 458, 1964. 

3. Fedorchenko, V. D. , Muratov, V. I., Rutkevich, B. N. Zhurnal Teoreti- 

cheskoy Fiziki, 34, 463, 1964. 



DISSIPATION OF PLASMA OSCILLATIONS EXCITED IN A 
CURRENT-CARRYING PLASMA 

Ye. A. Sukhomlin, V. A. Suprunenko, N. I. Reva, V. T. Tolok 

Several experimental and theoretical investigations have studied the 
development of bunched instabilities in a current- carrying plasma for 
large electric field strengths (Ref. 1-5). It has been shown that, as 
only the mean energy of the ordered electron drift is larger than their 
thermal energy, intense longitudinal electrostatic oscillations develop /127 
in a plasma. Their energy reaches the initial energy level of electron 
drift usually after several tens of plasma oscillation periods. 

Computations of the multi-flux motion of electrons in a current- 
carrying plasma, which were performed by 0. Buneman (Ref. 6), J. Dawson 
(Ref. 7), and Ya. B. Faynberg (Ref. 8), have shown that very intense 
"thermalization" of the plasma oscillation energy occurs, if this energy 
is considerably greater than the electron thermal energy. Thermalization 
occurs due to nonlinear phenomena leading to the transformation of longi- 
tudinal oscillations into transverse oscillations and to their rapid phase 
mixing . 

This process takes place until the energy of the ordered oscillations 
equals the electron thermal energy. It may be assumed that ion heating 

119 



will occur due to "collective" friction of electrons on ions in the case 
of bunched instabilities. 

Thus, an investigation of bunched instabilities in a high-current 
gas discharge opens up new possibilities for effective plasma heating. 

The studies (Ref. 9, 10) performed detailed investigations of the 
excitation conditions of bunched instabilities in a current-carrying 
plasma, as well as the plasma characteristics in the presence of these 
instabilities. The occurrence of an anomalously high discharge resistance 
and intense microwave plasma radiation was discovered. 

This article investigates heating and containment of a plasma in a 
strong magnetic field, under conditions when bunched instabilities excited 
by "escaping" electrons develop in the plasma. The experiments were per- 
formed on an apparatus representing a rectilinear tube made of alundvim 
with a diameter of 10 cm and a length of 25 cm, which was usually filled 
with hydrogen at a pressure of 5'10~^ - lO"** mm Hg. Aluminum electrodes 
were placed at the two ends of the tube; a battery of capacitors having 
an over-all capacitance of 15 microfarads was discharged between the 
electrodes. The battery was charged to a voltage of 30-40 kv. The 
discharge current through the gas amounted to 100 ka with a period of 

9 microseconds. In order to eliminate hydromagnetic phenomena, the dis- 
charge was performed in a strong longitudinal magnetic field (on the order 
of 1.2 tl) , at which the Shafranov condition of stability would be ful- 
filled (Ref. 13). In order that the plasma did not touch the walls, a 
diaphragm with an opening which was 80 mm in diameter was placed between 
the electrodes. 

During the first half-period in the discharge, a highly ionized plasm a/128 
filament, which was separated from the wall and which had a diameter of 
80 mm, was produced; no macroscale hydromagnetic Instabilities were 
apparent in this plasma filament. The plasma density changed between 

10 - 10^^ cm~^. The construction of the apparatus and the experimental 
method were described in detail in (Ref. 9). X-ray and microwave radia- 
tion from the discharge, the current of "escaping" electrons, the over-all 
discharge current, and the voltage between the electrodes were studied 
experimentally. 

Figure 1 shows the following oscillograms : a — microwave radiation 
from the plasma; b — over-all discharge current; c — current of "escaping" 
electrons; d — voltage between the electrodes of the discharge tube re- 
duced to a single time scale. The oscillograms were recorded at an initial 
hydrogen pressure in the chamber of 2-10"^ mm Hg, a magnetic field strength 
of 0.64 tl, and a charge voltage of 34 kv. 

Characteristic, inter-correlated oscillations are observed at high 



120 




t jiaicrosec 



Figure 1 



electric field strengths ; these oscillations are due to the development 
of bunched instabilities. During the initial period, the field strength 
in the discharge center increases as the voltage wave penetrates the 
plasma. With an increase in the electric field strength in the discharge, 
accelerating processes begin to develop. A current of "escaping" electrons 
first appears due to the "tail" of the Maxwell distribution. However, as soon 
as the electric field strength in the plasma begins to exceed the criti- 
cal value, all of the electrons acquire a drift velocity which is greater 
than the thermal velocity, and bunched instabilities develop in the plasma, 
to which the occurrence of epithermal microwave radiation corresponds. 
In this case, a large portion of the directional drift energy of the 
electrons is transmitted to excitation of oscillations, and the current 
of "escaping" electrons sharply decreases. This leads to an increase /129 
in the effective plasma resistance and to a dip on the oscillogram for 
the over-all discharge current. 



The amount of energy contributed by the outer source to the buildup 
of longitudinal electron oscillations may be computed from the additional 
current at the moment an instability develops. For the case shown in 
Figure 1, this energy amounts to 10 kv per particle. The energy of these 
oscillations considerably exceeds the initial thermal energy equalling 
30 electron volts, which leads to effective thermalizatlon of plasma 
oscillations due to nonlinear phenomena. As a result, the random electron 
energy will equal, in order of magnitude, the energy of plasma oscilla- 
tions. Due to thermalizatlon, intense X-ray radiation occurs as a result 
of energetic electrons falling on the target. 



121 



.£■ 


k 


^ 


3 


0.5 


\ 


nT^"^ 


I. 


\. 


■^v^^^ 






^ 


V ^"^ 


7—- s> 






^^ 


1 

2 


n 




1 / 





0,2 Ofi S.HM 

Figure 2 



Since the transverse component of electron energy has increased con- 
siderably, and the velocity of directed drift has remained the same as 
previously, the condition for excitation of plasma bunched instabilities 
has been disturbed and the plasma has returned to the initial unexcited 
state. Figure 1 clearly illustrates two cycles of such oscillations with 
a period on the order of 1.5 microseconds. This period is probably deter- 
mined by the time required for electrons to change into a state of "escape". 
According to the computations of Dreicer (Ref. 11), this period equals, 
in order of magnitude, the time between two Coulomb collisions for an 
electric field strength which is greater than the critical strength. 
Assuming that 30% of the energy of plasma oscillations is "thermalized" 
(Tg = 3*10^ electronvolts) (Ref. 12), we find that for a plasma density 
of 7 '10-^^ cm~^ the time between two Coulomb collisions is 4.5 microseconds. 

The period in which the "heating" cycles are repeated depends on the 
initial gas pressure in the chamber. It increases considerably with a 
decrease in the plasma density. The effective electron temperature of the 
plasma must thus increase, since the total energy transmitted into the 
buildup of plasma oscillations from the outer source changes very little. 

Thus, it would be expected that intense electron heating occurs due 
to the development of bunched instabilities in the discharge. Direct 
measurements of the electron temperature are of great interest. 



The effective electron temperature was determined by the absorption 
of electron braking radiation in thin beryllium foils located in front 
of a scintillation crystal on the wall within the vacuum chamber. The 
plasma electrons falling on the foil-target are braked in the very thin 
surface layer. Their energy is transformed into braking X-ray radiation 
which, after partial absorption in the foil, falls on the crystal causing 
a flash of light. In order to determine the radiation hardness, without 
disturbing the vacuum it is possible to place beryllium foils having 
different thicknesses in front of the crystal. The light from the crystal 



/130 



122 




Figure 3 



is supplied to the photomultlpller by means of the wave guide. The signal 
from the photomultlpller through the cathode follower is supplied to the 
oscillograph amplifier. 

Figure 2 shows the dependence of the photomultlpller (PM) signal 
intensity on the thickness of the absorber foil. This dependence may 
be employed to determine the electron temperature for a specific form 
of the electron energy distribution function. Maxwell distribution, 
rectangular distribution with the width T^, and Drayvesten distribution 

lead to similar temperature values . This enables us to employ the curve 
shown in Figure 2 for a rough estimate of the electron temperature when 
the electron energy distribution function is not known precisely. 

The curves in Figure 2 were compiled under the assumption of Maxwell 
distribution for three temperatures: 1 — 1 kev; 2 — 2 kev; 3-3 kev. 
It can be seen that the experimental points correspond to a plasma elec- 
tron temperature on the order to 2 kev. 

Figure 3 presents the following oscillograms : a — current of 
"escaping" electrons ; b — X-ray radiation from the plasma due to braking 



123 




J f ' I I L_i 1 1. .*.-J _ 



0^10 



mxcrosec 



Figure 4 



of thermalized electrons in the foil. X-ray radiation arises simultaneously 
with the current , and continues for a long period of time after the 
braking of "escaping" electrons due to the energy of transverse motion. 
The maximum quanta energy of this radiation is about 15 kev. This points 
to the effective transformation of the energy of electron longitudinal os- 
cillations into the energy of transverse motion. In the absence of 
bunched instabilities, the electron temperature in the discharge is 30 
electronvolts . 



124 



Figure 4 presents the following oscillograms: a — current of /131 
"escaping" electrons; b — X-ray radiation; c — microwave radiation; 
d — light from the discharge center; and e — over-all discharge current. 
X-ray radiation arises simultaneously with intense epithermal microwave 
radiation at a frequency close to 



y<+<-]/"^. 



(wq — plasma electron frequency; tOg — electron cyclotron frequency). 
The power of this radiation is approximately four orders of magnitude 
greater than the power of thermal radiation from the plasma at an elec- 
tron temperature of 10** electronvolts. It was shown in (Ref. 10) that 
the radiation is related to longitudinal electron oscillations in the 
discharge. 

All of these statements indicate that X-ray radiation from discharge /132 
is related to the thennalization of bunched instability energy. 

The period of pair interaction for electrons having an energy on 
the order of 2 kev is considerably greater than the time of the process 
being studied. Therefore, we may assume that nonlinear phenomena during 
collective oscillations with a large amplitude play the main role in 
thermalization of longitudinal oscillations. In our case, the Larmor 
radius of hot electrons is a little less than the discharge chamber 
diameter (less than 1 mm). Therefore, X-ray radiation is primarily 
caused by electrons diffusing toward the walls across the magnetic field. 
When the X-ray radiation reaches a maximum (see Figure 4) , the lumines- 
cence in the discharge center sharply increases; this is determined pri- 
marily by admixtures dislodged from the walls by hot electrons. There is 
a simultaneous strong increase in the current on the boundary electro- 
static probe, which is executed in such a way that the current upon it is 
determined by the resistance of the discharge plasma across the magnetic 
field. Thus, there is a rapid cooling of the electrons, even 5-6 micro- 
seconds after the X-ray radiation has terminated. 

In the absence of pair collisions, strong diffusion and great con- /133 
ductivity across the magnetic field would not be expected. Anomalous 
diffusion can occur due to nonlinear phenomena for a large longitudinal 
oscillation amplitude. However, in the experiments described, the anoma- 
lous diffusion across the magnetic field may be explained by an increase 
in the gaskinetic plasma pressure as compared with the magnetic pressure, 
due to intense electron heating. The decrease in the X-ray radiation in- 
tensity with an increase in the magnetic field strength also points to 
this conclusion. 



125 



REFERENCES 

1. Budker, G. I. In the Book: Material From the Conference on the 

Peaceful Utilization of Atomic Energy (Materialy konferentsii po 
mirnomu ispol'zovaniyu atomnoy energii) . Geneva, 4, 76, 1956. 

2. Thomassen, K. Phys. Rev. Letters, 10, 80, 1963. 

3. Shepherd, L. J., Skarsgard, H. M. Phys. Rev. Letters, 10, 4, 121- 

123, 1963. 

4. Zavoyskiy, Ye. K. Atomnaya Energiya, 14, 57, 1963. 

5. Fanchenko, S. D. , Demidov, B. A., Yelagin, N. I., Ryutov, D. D. 

Zhurnal Eksperimental'noy i Teoreticheskoy Fiziki, 46, 497 - 500, 
1964. 

6. Buneman, 0. Phys. Rev., 115, 503, 1959. 

7. Dawson, J. Phys. Ref., 113, 383, 1959. 

8. Faynberg, Ya. B. Atomnaya Energiya, 11, 313, 1961. 

9. Suprunenko, V. A., Faynberg, Ya. B., Tolok, V. T., Sukhomlin, Ye. A., 

Reva, N. I., Burchenko, P. Ya. , Rudnev, N. I., Volkov, Ye. D, 
Atomnaya Energiya, 14, 349, 1963. 

10. Suprunenko, V. A., Sukhomlin, Ye. A., Reva, N. I. Atomnaya Energiya, 

17, 83, 1964. 

11. Dreicer, H. Phys. Rev., 115, 238, 1959. 

12. Shapiro, V. D. Zhurnal Eksperimental'noy i Teoreticheskoy Fiziki, 

44, 613, 1963. 

13. Shafranov, V. D. Atomnaya Energiya, 5, 38, 1956. 



126 



■■ III mil I mil HHHi mil i i ii i i n 



I 



DAMPING OF INITIAL PERTURBATION AND STEADY FLUCTUATIONS 
IN A COLLISIONLESS PLASMA 

A. I. Akhiyezer, I. A. Akhiyezer, R. V. Polovin 

When the perturbation of the distribution ftinction is a nonanalyti- 
cal function of velocity, the damping of plasma waves when there are no 
collisions may be different from that determined by the classical formula 
of Landau. If the perturbation of the distribution function has a /134 
6-like singularity, then perturbations of the macroscopic quantities are 
not damped with the passage of time. If the perturbation of the distribu- 
tion function has a discontinuity of the n-th derivative, then the perturba- 
tion is damped according to the law t-(n+l) (and not according to the 
exponential law, as in the case of Landau damping). 

Let us investigate the mechanism for establishing fluctuations 
(which are not dependent on the initial perturbations) of the macroscopic 
quantities in a non-equilibrium plasma. We shall show that these fluctua- 
tions are established due to the "survival" of a singular component in 
the expression for the distribution function perturbation. 

The dependence on time of the perturbation of the k-th component of 
the Fourier potential <t>i^(t) in an unconfined plasma is determined by the 

following expression (Ref. 1) 

f+'^o (1) 

O — ioo 



'^^^^^ _ 4^ Af (k, p) . 

?Ap - k^ ' D (k, p)' 



(2) 



N{k,p)=jjk{w)~^; (^j 



w=^ ,F^iw) = ^Fo{v)dvj., gk(w)^^gki\)dv^, vx=v--k|; 



Fo(v) is the unperturbed distribution function; g^Cv) — the Fourier com- 
ponent of the distribution function perturbation in the case of t = 0; 
integration is performed in formula (1) along the line Re p = a lying to 
the right of all the singularities of the function i> . 

Formula (1) makes it possible to determine the behavior of the poten- 
tial <f>]j^(t) with an increase in t. As is known, the asymptotic behavior 

127 



of the function <i)]j_(t) for large t is determined by the nature of the 

singularities of the Laplace transform (J)^. The function (|)p was 

detemnined above for only large values of Re p (in this region it has 
no singularities). In order to study its singularities, we must first 
determine this function in the entire complex variable plane — i.e., 
we must analytically continue the determination of (2) to decreasing 
values of Re p. The analytical continuation of (jip is determined according 

to the previous formula (2) along the imaginary axis p. 

For purely imaginary values of p, the denominators in the integrals /135 
which determine (j)„ vanish in the case of w = ip/k. Therefore, for 

analytical continuation of (j) in the region Re p ^ 0, it is necessary 

to deform the integration path in the integrals (3) and (4) , so that it 
passes around the pole w = ip/k from below. Deformation of the path 
assumes, in its turn, the possibility of analytical continuation of the 
functions F^ (w) and gjj^(w) determined initially only for real w in the 

region of complex values of w. 

Thus, a clarification of the singularitites for the function ((Jcr* 
which determines the nature of the asymptotic behavior of i|)]j^(t) for large 
t, requires a knowledge of the analytical properties of the functions 
Fo(w) and g^Cw). 

Let us confine ourselves to investigating the functions Fq (w) per- 
mitting analytical continuation in the region of complex values. The 
function D(k, p) , determined in the case of Re p > by relationship (3), 
may be continued analytically in the region Re p .< 0, by determining it 
everywhere as 



'x^.")"'-^^!!!^' « 



-0' + 

where integration is performed along the real axis w with passage around 
the pole from below in the case of w = ip/k. 

We have found the denominator of expression (2) for * — i.e., the 

function D(k, p) — over the entire plane of the complex variable p. Let 
us now calculate the analytical properties of the numerator for this expres- 
sion, i.e., the function N(k, p) . Formula (4) determines it in the case 
of Re p > 0. As has been pointed out, the fimction N(k, p) has no singu- 
larities in this region. The position and nature of the singularities of 
this function are determined by the properties of the fvinction gjj^(w) in 

the case of Re p ^ 0. 

If the function gj^(w) has singularities (which may be integrated) for 

128 



real w, then the function N(k, p) will have singularities for purely 
imaginary p. In particular, such a situation is observed if the func- 
tion g]5^(w) has a 6-like singularity, a discontinuity, or a break, and 
also if any of its derivatives has a break (in these cases, the func- 
tion gjj^Cw), generally speaking, does not permit analytical continuation 

on the real axis. 

If the function g]^(w) has no singularities on the real axis and per- 
mits analytical continuation in the region of complex values of w, then 
the function N(k, p) , and consequently the function (j) , will have no 

singularities on the imaginary axis p. However, generally speaking, it /136 
may have a singularity in the case of Re p < at the points p = -ikw^., 

where Wj. is any singularity of the function g^^Cw) lying in the lower 

halfplane of the complex variable w. 

Let us elaborate further on the nature of the asymptotic behavior of 
(t)jj^(t) for noninteger ftmctions g^(w) . In this case, singularities of the 

function N(k, p) are added to the singularity ^ determining the roots of 

the dispersion equation D(k, p) = 0. The distribution of these singulari- 
ties depends only on the form of the function gu(w) — i.e., on the nature 

of the Initial perturbation — and does not depend on the plasma properties 
(on the function Fg (w) . ) As has been indicated, one significant property 
of the singularities for the function of N is the fact that they may all 
lie only in the left halfplane p. Therefore, if only one of the roots 
Pj. = -ia)j,-Yr of the dispersion equation D(k, p) = lies in the right 
halfplane p, Yr *^ (which corresponds to the possibility of an oscilla- 
tion increase) , then the nature of the initial perturbation has no signifi- 
cant influence on the asymptotic behavior of <l'i^(t) in the case t t ->■ °°. 

If N(k, p) has singularities at the points p = p^^ = -Yn~i<J^n (ii = 1> 
2, 3, ...), the contribution made by these singularities to the asymptotic 
behavior of <t'k(t) in the case of t ->- «> may be written as2j""®^P{ — T«^ — "o«0, 

where a.^ represents certain constants. Adding this svim to the contribu- 
tion from zeros D(k, p) , we find the asjnnptotic expression for 'f']j^(t) in 
the general case of noninteger functions g]j^(w) (which have no singulari- 
ties for real w) 

"fx (0 ~ S?!'-^ exp { —i^i - /CO,/} + ^ a„ exp {—^J - /co„/}, ^^^ 

where <\>^^^ is the residue (Ji^,^ at zero of the function D(k, p) (the point 

cr 
P = Pr = -Yr-if^r^* 

129 



Thus, for large t the potential <i>v(t) represents superposition of 
the eigen plasma oscillations, whose complex frequecies a)j.-iY^ are deter- 
mined by the plasma properties [the right sum in (6) ] , and oscillations 
whose complex frequencies (jJ^-iY are determined by the form of the initial 

perturbation gi^(w) [the second sum in (6)], The eigen oscillations may 

be both damped and intensified. The oscillations whose frequencies are 
determined by the form of the function gi^(w) may be only nonincreasing 

(i.e., damped or oscillating oscillations with constant amplitude). 

Let us give two examples of oscillations whose frequency and damping 
decrement are determined by the initial perturbation, and do not depend 
on the plasma properties. 

As the first example, let us investigate oscillations produced in /137 
the case of 



gkiw) 



go^i 



(w-w,y+w^i' O) 



where gg , wq , wj are certain constants. In this case we have 

N (k, p) = , ■■ "\ . . 

The function N(k, p) has a singularity in the case of p = -ikwQ-kwi, 

which introduces the following contribution to the asymptotic behavior 
of <l'v(t) in the case of t -^ ~ 

fk (0 "^ ^0 exp [—kwif — ikw^t]. (8) 

Thus, the frequency and damping decrement of oscillations produced in 
the case of initial perturbation such as (7) equal kwQ and kwj_, respec- 
tively. In the case of wj -^ 0, the damping disappears. We should point 
out that the function gi^(w) acquires an 6-like singularity on the real 
axis, g]^(w) -^ Trgo6(w-wo). 

Let us study oscillations produced in the case of the discontinuous 
functions g, (w). Let us set, for example, 

^0 (—^0 <W< Wo), (9) 



Thus, the function 



^^(^)-lO {\w\>w,) 



130 



^ '^' IK p — ikW 

has branch points on the Imaginary axis p, p = + ikw. The contribu- 
tion made by the singularities of the function N(k, p) to the perturba- 
tion of the potential (("kCt) has the following form 

^,(0~^/iiii«. (10) 

The 6-like singularity on the real axis on the function g}^(w) leads to 
nondamped oscillations of the potential 'l'ir(t) • The discontinuity of the 
function gj^Cw) — i.e., the 6-like singularity of its first derivative — 
leads to potential oscillations which are damped as t~^ . It can readily 
be shown that the discontinuity of the n-th derivative of the function 
g]j^(w) — i.e., the 6-like singularity of its (n + l)-th derivative — 
leads to asymptotic behavior of a t~(^'''-'-)exp{ikwot} like potential, where 
wg is the discontinuity point. 

Let us determine the manner in which the fluctuations of macroscopic /138 
quantities, which do not depend on the initial conditions, are established 
in a collisionless plasma with an arbitrary (not necessarily equilibrium) 
distribution function (Ref. 2). In order to do this, we should note that 
the averaged product of the distribution function fluctuations for particles 
of the a-th type can be represented as follows at corresponding periods 
of time 

< 8t (V) gt- (V) > = iaa'^ (k + k') 8 (V - V') fS (V) -f Yaa' (v, v'; k, k'). ^^^^ 

where the first component (Ref. 3) describes the "correlation of the particle 
with itself" , and the second component is related to the interaction between 
particles (and is determined by the previous history of the system) . It 
is important that the first component contain 6(v - v'), while the second 
component is a smooth function of velocity. 

In order to obtain the correlation function of the potential, we 
must express the potential <t']j(t) by gg(v) by means of relationships (1) - 

(4) , and we must then perform averaging by means of formula (11) . It may 
be readily seen that the presence of the 6-like component in formula (11) 
leads to a nondamped (and oscillating according to the law expdt^yfc}) com- 
ponent in the expression for the potential correlator. The remaining 
(smooth) components in formula (11) are damped according to the law 
exp{-Yi^t}, where Yi, is the customary Landau damping decrement (i.e., the 

imaginary part of the root ny, of the dispersion equation e(a), , k) =0). 

131 



Thus, in the time yy~ the potential fluctuation distribution (and 

the distribution of all other macroscopic quantities) , which does not 
depend on the initial perturbation, is established after the outer per- 
turbation is shut off in the plasma. 

According to relationships (1) - (4) , (11) , the correlation function 
of the potential has the following form 

-< n (0 <PA'(0) > = 8 (k + ki^Y r do) I s (to.k) |-2 exp { - mt) x 

^ -i (12) 

X2e2^dv/=?(v)8(u>-kv). 

a 

This relationship was obtained by Rostoker (Ref . 4) by employing a 
different method. Our derivation presents a clearer explanation of the 
mechanism, and makes it possible to determine the time required to estab- 
lish fluctuations, which do not depend on the initial perturbation, in a 
nonequilibrium plasma. 



REFERENCES 

1. Landau, L. D. Zhumal Eksperimental'noy i Teoreticheskoy Fizikl, 16, /139 

574, 1946. 

2. Akhiyezer, I. A. Zhumal Eksperimental'noy i Teoreticheskoy Fiziki, 

42, 584, 1962. 

3. Kadomtsev, B. B. Zhumal Eksperimental'noy i Teoreticheskoy Fiziki, 

32, 943, 1957. 

4. Rostoker, N. Yademyy Sintez, 1, 101, 1961. 



132 



CHARGED PARTICLE INTERACTION WITH A TURBULENT PLASMA 

I. A. Akhlyezer 

This article computes the energy lost (or acquired) per unit of time 
by a charged particle when moving through a turbulent plasma. The depen- 
dence of the particle energy change on the magnitude and direction of its 
velocity is established in explicit form. It is shown that the turbu- 
lence spectrum elements have no influence on this dependence. 

The strength of charged particle interaction with the plasma is deter- 
mined by the level of plasma fluctuations. In particular, the particle 
energy losses per unit of time P are related to the charge density corre- 
lator <P^>qtQ by the following relationship 

" = - '-f" I(?)'» < '' > '-« (•-''- "f ) fw • "> 

where ez, y, v are, respectively, the particle charge, mass, and velocity. 
If the plasma consists of cold ions and hot electrons moving at the mean 
velocity u with respect to the ions, the charge density correlator in 
the "sound region" (q(T^/M) ^2 << ^ << q(T /m) ^2^ aq << i) may be repre- 
sented in the following form 

( P' > ,. = f 7' {aqr {[t {q, qu) - ^ "-^(o] 8 (co - qs) + 
+ [t {q, -qu) - i-'^u)] S (u, + gs)] , 

where T„, T- is the temperature and m, M — the mass, respectively, of 
electrons and ions; s = (T„/M) ^^ — speed of sound; T(q, qu) — effective 

1 1 

temperature of sound waves; a = T ' 2(4Tre^n)~ '^ — Debye radius. Substi- /lAO 
tuting (2) in (1) and assimiing that T >> pv^ , we obtain 

(eza)^us . 



P = 



' ^g'dq f dcp jcos 9 - (^ - l) '%in 6 cos f\§-^T {q, tj). (3) 



where 6 is the angle between v and u 

Tj^qu = ^u |-cos9 + (1 — ^1 sin9cos<p|. 

In the wave vector region in which the sound waves are damped (qu < 
< qs) , the effective temperature is (Ref. 1-3) Tg(l - qu/qs)~^. Close 

to the boundary of the stability region (qu ^ qs) the function T increases 



133 



sharply. In the case of qu > qs, the linear theory predicts an exponen- 
tial increase in T with time. This increase stops due to nonlinear phe- 
nomena, and a stationary fluctuation distribution is established \rtiich 
is characterized by a very high effective temperature, i.e., the state 
of stationary turbulence (Ref. 4). 

It follows from this behavior of the ftinction T(q,Ti) that 3T(q,n)/3n 
has a sharp maximum for a certain value of n close to qs, n = no '^ qs. 
Noting that the energy losses are determined by the derivative 8T/3n, 
and not by the function T itself, we can thus express P by a small number 
of parameters which characterize 9T/8n in the case of n ^ tio» without 
Including more detailed properties of the turbulence spectrum. 

Close to n '^ no» we have 

{,^r(,..)}-. =,^pyM(.) + xi(,)(i -fj}, (4) 

where \i 2 equals unity in order of magnitude, and T* is a large quantity 
equalling T(q, n) in order of magnitude in the case of n > qs. 

Substituting (4) in formula (3) and assuming that 6^ < 6 < e_, where 

cos e± = (ay)-' {s^ ± (v^ — s^^'^w" — s^)'/.}, (5) 

we obtain 

P = ^-^S-' '^ ^^ (^°s ^+ - COS ^y^'' (COS - cos 9_)-'A, (6) 

where a = afiaq)^ {\i\2)~^^'i- This relationship determines the dependence 
of P on V in explicit form. In the case of 6 < 60(60 = arc cos s^/uv) , 
the particle energy decreases and in the case of 6 > 60 it increases. 

In the case of |6 - 6^. ]^ Tg/T*, relationship (6) ceases to be valid. /ML. 
In the case of | 6 - 6+ | << T^/T*, we have 

-aV("fT,sin9^//' IV / V / • (7) 

where a^ =^aliaq)^\T'^'\V'^'dq (the signs «+» correspond to 6 ;^ 6^ ) . 
Thus, P is proportional to T* in the region 6, < 6 < 6_, except for the 
boundary of this region where P "^ (T*) '2. 

In the case of v -^ u, the critical value of the angle 6^. strives to 



134 



zero, and relationship (7) does not hold. For 6^. << T^/T*, and employing 
formulas (3) , (4) , we obtain 

In this case, the energy losses are particularly large (they are propor- 
tional to (T*2) . 

If < 6^. or 6 > e_, then the expression for the energy loss does 
not contain the large parameter T*/Tg. For | - 6^ | « 1, nevertheless, 
P is proportional to |6 - 6^. |~ /z, and consequently it is large. 

In order to have critical values of the angles 0^ , it is necessary 
that both u and v exceed s. If u ^ s , then 0. ^ = arc cos s/v. In 



± "c 



the case of ol 0c» the energy losses are propotional to (T*) /2 

When expressions (6) - (9) were derived, it was assumed that the 

difference v - s was not too small. If 1 - s/v << T^/T*, the energy 

losses will be at a maximum in the case of vu = s^ (in this case P is 
determined by formula (8) and sharply decreases with an increase in 
|vu - s^l . 

In conclusion, we would like to point out that the dependence of P 
on the angle holds, even if the assumption that 9T/3n is small in the 
region n > no is not fulfilled. The contribution made by the quantity 
9T/8n with n > Ho in the expression for P can only change the function P 
somewhat in the case of 6^ < < 6_, without changing P in the case of 
0^0^. Consequently, the nature of the dependence of P on the angles 
is not changed in the case of 0^ <: .< 0_. In particular, the function 
P, which is positive in the case of = 0+ and negative in the case 
of = 0_, must vanish for a certain value of the angle = Oq, 6+ < 
< 0Q < 0_ (thus 00 can differ somewhat from arc cos s^/uv) . 

REFERENCES /1A2 

1. Ichimaru, S., Pines, D. , Rostoker, N. Phys. Rev. Letter., 8, 231, 

1962. 

2. Ichimaru, S. Ann. of Phys., 20, 78, 1962. 

135 



3. Bogdankevich , L. S., Rukhadze, A. A., Silin, V. P. Izvestlya Vuzov. 

Radiofizika, 5, 1093, 1962. 

4. Kadomtsev, B. B., Petvlashvlli, V. I. Zhurnal Eksperimental'noy i 

Teoretlcheskoy Fiziki, 43, 2234, 1962. 



THEORY OF NONLINEAR MOTIONS OF A NONEQUILIBRIUM PLASMA 

I. A. Akhiyezer 

As is well known, low frequency oscillations with a linear law of 
dispersion — so-called ion sound — are possible in a collisionless 
plasma consisting of hot electrons and cold ions (Ref. 1, 2). It is 
interesting to study nonlinear motions of a collisionless plasma consisting 
of hot electrons and cold ions, and primarily simple waves. The study of 
simple waves not only makes it possible to trace the development of an 
initial perturbation, but it is also of interest as an independent investi- 
gation, since only the region of simple waves can (when there are no dis- 
continuities) be contiguous to an unperturbed plasma [see (Ref. 3)]. 

A. A. Vedenov, Ye. P. Belikhov, and R. Z. Sagdeyev (Ref. 4) have 
studied simple waves in a two-temperature plasma on the basis of an ios- 
thermal hydrodynamic model. This article investigates simple waves in a 
nonequilibrium plasma on the basis of a kinetic equation, without employing 
a special model. A system of equations has been obtained for the moments 
(introduced in a specific way) of the electron distribution function, which 
enabled us to determine the direction of change for quantities characterizing 
the plasma in a sound wave, and to trace the development of a perturbation 
having finite amplitude*. 

» Equations Describing a Simple Wave /143 

The system of equations describing the motion of a collisionless 



Yu. L. Klimontovich and V. P. Silin (Ref. 5) have investigated the 
problem of a hydrodynamic description of a two-temperature plasma 
without collisions. Nonlinear motions of such a plasma were studied 
in (Ref. 4, 6). 



136 



plasma consisting of hot electrons and cold ions has the following form 

(l + "l)" + :&E = 0; |^n + div(«u) = 0; (D 

div E = iTze (^Fdv — n) ; rot E == 0, 

where F(v) is the electron distribution function; n and u are, respectively, 
the ion density and hydrodynamic velocity; E — electric field; m, M — 
electron and ion masses, respectively. (It is assumed that the electron 
mean energy considerably exceeds the ion mean energy.) Being interested 
in sound oscillations whose phase velocity is small as compared with the 
mean thermal velocity of electrons, we do not have to take the term 8F/3t 
into account in the first of the equations (1) . Confining ourselves to 
one-dimensional plasma motions and making allowance for the fact that the 
charge spatial distribution is small in a soundwave 

(X — the length at which the quantities characterizing the plasma change 
significantly; a — Debye radius), we can reduce the system of equations 
(1) to the following form 

dF __M du dF _ f,. ^ . „ ^ _ n- 

dx m ' dt ' v^dv^ ~ ' dt '^ dx — ^' (2) 

n — ^ Fd\, 

where d/dt = 8/9t + u^^x (the x axis is selected in the direction of wave 
propagation; the subscript x for the velocity component u^^ is omitted from 
this point on) . 

Let us introduce the "moments of the distribution function": 

D; (X. t) = (-2)' r ^ F (v; x, t)dv, / = 0. 1 ^3^ 

Employing system (2) , we obtain 

SDj M du ^ _. dn , au ^ _ (4) 

In order to study nonlinear plasma motions, system (4) is more advan- /144 
tageous than the initial system of equations (2) , since it includes tenns 
which are only dependent on x and t, while equations (2) also include the 
electron velocity distributions. In this sense, equations (4) are similar 



137 



to equations of hydrodynamics, although — in contrast to hydrodynamics 
which operate with a finite nimiber of quantities — they Include an 
infinite number of "hydrodynamic quantities" n, D- , u. 

We are Interested in simple waves, i.e., those plasma motions for 
which the perturbations of all quantities characterizing the plasma are 
propagated at the same velocity — in other words, for which each of the 
functions X [X = u, n, D . , F(v)] satisfies the following equation 

Id . ,,, .^ a 



|+K(^,0|)x = 0. 



In the case of simple waves, as is well known, all of the quantities X 
can be represented in the form of a function of one of them (for example, 
n) , which in its turn is a function of x, t. System (4) thus changes into 
a system of customary differential equations for the functions Dj (n) , u(n), 

and the phase velocity V(n) is determined from the solvability condition 
of this system. After simple transformations, we obtain 

du ^ !j . '^^ -^ . 

dn ^ n ' dn ^ Di ' (5) 

where e = +1 (e = -1) , if the wave is propagated in the positive (negative) 
direction of the x axis. 

The determination of the electron distribution function in the case of 
a simple wave may also be reduced to solving the customary differential 
equation. Rewriting the kinetic equation (2) in the following form 

dF{v) M _ (V ~uY dF{y) _ 
dn ' m n ^x^^x 

and introducing the notation 

F[vl; vr, n)=BF{v; n) 

[v^. = (Vy, Vg.)], we obtain 

where the function 3(n) satisfies equation /145 



dp , 2IVI V\ f. 

- — . — = u. 

an. m n 



138 



(7) 



We should point out that Landau damping of sound waves was not 
taken into account when the initial equations (2) were derived. There- 
fore, it is necessary that the wave amplitude An not be too small, 
An/n >> (m/M) /2, in order that equations (2), and consequently relation- 
ships (4) - (7) may be valid. The role of nonlinear phenomena in the 
development of the perturbation is much greater in fulfilling this condi- 
tion, than is the role of sound damping. 

Development of a Perturba tion Having a Finite Amplitude 

The system of equations for "hydrodynamic" quantities (5) , together 
with relationships (6) and (7) , enables us to study the direction of the 
change in quantities characterizing the plasma (including the electron 
distribution function) and to trace the development of a perturbation 
having a finite amplitude. 

First of all, let us determine the manner in which the electron dis- 
tribution function changes in a simple wave. It follows from (7) that 3 
decreases in a contraction wave, and increases in a rarefaction wave. 
Therefore, for values of v for which d F/v^ v^ < 0, the number of electrons 

having a velocity in the (v, v + dv) range increases in the contraction 
wave, and decreases in the rarefaction wave. Conversely, at values of v 
for which3F/Vj^v > 0, the number of electrons with velocities in the 

(v, v + dv) range increases in a rarefaction wave, and decreases in a con- 
traction wave. In particular, if the initial electron velocity distribu- 
tion has a spike encompassing a small velocity region, along with a maximum 
for Vjj = 0, the spike shifts to the region of larger (smaller) values of 

|vjj| as the contraction wave (rarefaction) moves. 

Let us dwell in somewhat greater detail on the case of Maxwell distri- 
bution. Employing system (5), we may state that in this case Vg, D./n do 

not depend on n, and consequently are motion integrals. The electron tem- 
perature and the distribution function F/n, which is normalized to one 
particle, do not change during wave propagation. Thus, in the case of 
a Maxwell velocity distribution of electrons it is valid to describe a 
two-temperature plasma by means of isothermic hydrodynamics . 

In order to determine the manner in which the form of the sound wave /146 
changes, it is necessary to compute the derivative dV/dn [see (Ref. 7)]. 
Employing the system of equations (5), and asstmiing, for purposes of 
definition, that e = 1, we obtain 



139 



Depending on the electron distribution function, dV/dn may be positive, 
negative, equal to zero, or alternating [positive for single values of 
the parameter g and negative for other values of this parameter, see 
formula (6) ] . (We should point out that the derivative dV/dn is always 
positive both in customary and in magnetic hydrodynamics.) 

If dV/dn > for all values of the parameter g, then (just as in 
customary hydrodynamics) points with a large density move at a large velo- 
city. Therefore, discontinuities arise at the contraction sections.** 
Self-similar waves are rarefaction waves. In particular, this possibility 
exists for a Maxwell electron velocity distribution and for a distribution 
in the form of a step F -v (vQ(n) - v^) , G (x) = -^ (1 + sign x) . 

If for all values of g, dV/dn = 0, all the points move at the same 
velocity during wave propagation. Therefore, the wave profile is not 
deformed and no discontinuities arise. Employing equations (5) and (7), 
we may state that the velocity of two-temperature sound and the quantity 
B '2 change in an inverse proportion to density, V n = const, Bn^ = const. 

The case dV/dn = is realized, in particular, for a Cauchy distribution 

F "^ {vo(n) + v2}-2. 

If dV/dn < (independently of the value for the parameter g) , then 
points with a large density move at a low velocity. Therefore, discon- 
tinuities arise in the rarefaction sections. Self-similar waves are con- 
tractions waves. This possibility is realized, in particular, if the dis- 
tribution function is the superposition of two Cauchy distributions 

F- V, («) [v' + v', (n)}-'+ V2 («) [v' + vl {n)]-\ 

We should note that in this case the velocity V increases in the rarefac- /^^^ 
tion wave, and decreases in the contraction wave. 

Finally, let us discuss the case when dV/dn may be both positive and 
negative, depending on the value of the parameter g. For purposes of 
definition, we shall assume that dV/dn > in the case of g > gj^ and 
dV/dn < in the case of g < gi, where gi is a certain critical value of 
the parameter g. When a contraction wave moves in such a plasma, the 



** We have employed the term "discontinuity" to designate the narrow 
regions in which the gradients of the quantities characterizing the 
plasma become so large that the initial equations (2) are not 
applicable. In the case of a/X ^1, sound dispersion must be taken 
into account. With a further increase in the gradients, multi-flux 
flows [see (Ref. 4)] or Shockwaves may arise in these regions. 



140 



"apex" of the wave (points with the density n > nj , where n^ is deter- 
mined from the equation 6(ni) = 3i) lags behind the "base" of the wave 
(points with n < nj) . Therefore, the density at the point of the dis- 
continuity which develops from such a wave cannot exceed nj. A discon- 
tinuity may also arise during the motion of a rarefaction wave. Thus, 
the density cannot be less than n^ at the discontinuity point. If, on 
the other hand, dV/dn < in the case of B > 62 and dV/dn > in the 
case of 3 < 32» then — as may be readily confirmed — the density at 
the discontinuity point cannot exceed n2 when a discontinuity develops 
from a rarefaction wave, and cannot be less than n2 when a discontinuity 
develops from a contraction wave (the critical density n2 is determined 
from the equation 3(n2) = 32)' 

Both of the above possibilities may be realized, in particular, if 
the electron velocity distribution is a superposition of two Maxwell dis- 
tributions, "hot" and "cold", 

F{v; n) = vi (n) exp { - ^ j + v^ (n) exp { - ^'} . T^ » T,. 

For small density values (n<^{aiT'i'/a2T'2'Y'^^'), according to equations (6), 
(7), Vi = ai/i; V2 = a2«^"'^' («!. ^2 — constants). It may be readily confirmed 
that in this case dV/dn > in the case of n < nj and dV/dn < in the case 

of n > ni , where «, = (H^ . ^) . For large density values {if^ a'iT'i'/a'2T'2') 

\ °2 t/'J 

Vj = airt^'^'^'; V2 = a2« («!. 12 — constants). In this case, dV/dn < for 

n < n2 and dV/dn > in the case of n > n2, where n2 = —.{TjTi)'''. 

2a 2 

Let us investigate the motion of a two-temperature plasma arising 
during its uniform contraction or expansion [similarly to the problem of 
the plunger in hydrodynamics, see the monograph (Ref. 7)]. We shall 
assume that the plasma occupies the half space x > Vot, which is uniformly /148 

limited by a moving plane. (Such a boundary may represent, in particular, 
the region of a very strong magnetic field.) As is well known, only self- 
similar waves (in the absence of shock waves) can be steady motions of 
a uniformly contracted (or expanding) medium. If dV/dn > 0, a self- 
similar wave (which is in this case a rarefaction wave) arises during 
plasma expansion (Vg < 0) . If dV/dn < 0, a self-similar wave (which is 
in this case a contraction wave) arises during plasma contraction (Vq > 0) . 

Employing formulas (5) , (6) , (7) , we may relate the change in all 
the quantities X characterizing the plasma in a self-similar wave with the 
"plunger" velocity Vg. Assuming, for purposes of simplicity, that 

141 



I'n. .r. 0/+I Vo. 



s 



Vq « V , we obtain 

REFERENCES 

1. Tonks, L., Langmulr, I. Phys. Rev., 33, 195, 1929. 

2. Gordeyev, G. V. Zhurnal Eksperimental'noy i Teoreticheskoy Fiziki, 

27, 18, 1954. 

3. Akhiyezer, A. I., Lyubarskiy, G. Ya. , Polovin, R. V. Zhurnal 

Teoreticheskoy Fiziki, 29, 933, 1959. 

4. Bedenov, A. A., Velikhov, Ye. P., Sagdeyev, R. Z. Yadernjry Sintez, 

1, 82, 1961. 

5. Klimontovich , Yu. L. , Silin, V. P. Zhurnal Eksperimental'noy i 

Teoreticheskoy Fiziki, 40, 1213, 1961. 

6. Nekrasov, F. M. Zhurnal Teoreticheskoy Fiziki, 32, 663, 1962. 

7. Landau, L. D., Lifshits, Ye. M. Mechanics of Solid Media (Mekhanika 

sploshnykh sred) . Gosudarstvennoye Izdatel'stvo Tekhnicheskoy i 
Teoreticheskoy Llteratury, Moscow, 1953. 



142 



NONLINEAR PROCESSES IN A UNIFORM AND ONE-COMPONENT PLASMA 

N. A. Khizhnyak, A. M. Korsunskly 

Nonlinear solutions of a one-dimensional kinetic equation without a 
collision term, which depend on x and t by the combination ? = x - Vot, 
where — Vq is the constant wave velocity, were compiled in (Ref . 1) /149 
and studied for several cases in (Ref. 1-4). The conditions at which 
these solutions may be realized were found in (Ref. 4), and a limiting 
transition to small oscillations was performed. 

General Theory of Nonlinear Waves 

This article investigates the more general nonlinear solutions of 
a one-dimensional kinetic equation without collisions 

dt ^ "■ dx dx du — "' 

where <}) is the potential of the self-consistent electric field with the 
factor — , determined by the Poisson equation 

where ng is the unperturbed density of ions whose mass is assumed to be 
infinitely large. It is assumed that the distribution function f depends 
only on the variables u and ^. Then the electron density 

< n ) =^fdu==F(<f), (3) 



the density of the electron fltix 

< nu ) = Ju/d« = 4>(9) 



(4) 



and the energy density 

(n^ > =J/(u)f-'d«=^(cp) (5) 

are explicit functions of only the potential (j). 

We shall show that in this case the main plasma characteristics may 
be compiled within the framework of a hydrodynamic approximation, and 
that the electron velocity distribution function may be found relatively 
simply. From the equation of continuity 



143 



we find that the potential (}> must satisfy the following equation 

|+l^o(T)g = 0. (6) 

where Vq = — ^ is a certain velocity of longitudinal waves in the plasma 

which depends on the potential (f). It can be readily seen that the specific /150 
solutions depending on x - Vot, where Vq = const, are special cases of 
equation (6) in the case of Vo(i)>) = const. 

From this point on, it will be assumed that the function Vo(<|)) is 
given, and it may be employed to formulate the solutions both for the 
equations of the hydrodynamic approximation and for the kinetic equation. 

The equations of the hydrodynamic approximation 

may be reduced to equation (6) and to the equation for hydrodynamic velo- 
city v(x, t). In actuality, since 



( n > 



it follows from expressions (3) and (4) that v = v((t)) . Therefore, the 
Navier-Stokes equation (Ref. 7) has the following form 



— r°!? -1- ^fl — ^1 I dp d<f 



and is identical to equation (6) , if only v(<|>) is determined by the equa- 
tion 



-^;+itizi.^v^(,). (8) 



¥ 



The electron plasma density <n> may be found from the specific wave 
velocity VgCtf') and the hydrodynamic velocity v((j)) from equation (8), by 
means of the following relationship 

-f "^ df 

„ , . J a (9) - I'. (V) 



144 



which follows from the equation of continuity. Thus, the dependence 
of the main hydrodjmamic plasma parameters on the potential (|i is always 
formulated in quadratures for a certain Vo(i))). 

The potential of the self -consistent field (^ is determined according 
to equations (6) and (2) . The general solution is found in the following 
form from equation (6) according to a certain function Vo('l') 

? = ? (c). 

where c = c(x, t) is the equation of its characteristic. The characteris- 
tics of a quasilinear equation in partial derivatives (6) may be com- 
piled according to the well known method (Ref. 5). 

We shall regard t, x and (j) as functions of a certain parameter s. /151 
Then the parametric equation of characteristics can be written as 
follows 



dt , dx ,/ .V. d<f 



from which it follows that <\> equals ^q and does not depend on s. There- 
fore, X = Vo(<(>)s + XQand t = s + tg, where Xq, tg and (J>o are the values of 
t, X and (f) on a certain line s = through which the characteristics pass. 
In particular, if primary Interest ia directed toward the development of 
moving waves, which is caused by the nonllnearity of the medium, xq, to 
and ^Q must characterize a given Initial moving wave. 

Let us assume that at the initial stage of the process there is a 
given moving wave with a constant phase velocity wq . We then have 

where ^q^i) characterizes the initial dependence of the field potential 
on time. In this case, we have 

/ = s + t; ^ = ^0(^0 (^)) « + K'o'^; <P = To i^) 
and after excluding s we find t from the following equation 

which determines the characteristics of the quasilinear equation (6) 
T = t(x, t) . Using the characteristics from equation (2), let us deter- 
mine the field potential in the plasma <()(x, t) and all of the hydrodynamlc 
parameters of the medium. 



145 



\% 



We can find the electron velocity distribution function from the 
kinetic equation (1). Actually, since equation (1) is transformed into 
the following form, together with equation (6) , 

[«-l^o(?)]|-| = 0. 
the equation of characteristics 

enables us to find the particle velocity distribution function in a 
general form from the potential (j) and the kinetic velocity u. Since the 
potential ^ may be compiled within the framework of the simpler hydrody- 
namic approximation, the specific dependence of the distribution function 
on the coordinates and time has been established. 

Propagation of Waves in Media_wit h Li near Dispersion /152 

We shall assume that the given law for the dependence Vq ((j)) of the 
nonlinear wave phase velocity on potential determines the law of the 
medium dispersion. We shall study nonlinear, nonstationary waves in a 
medium with linear dispersion 

where Vq and y are certain constant parameters. 

We shall disregard the plasma pressure, so that equation (8) assumes 
the following form 

■i = -^ + ^o + -(?- 

The dependence of the hydrodynamic velocity v of the plasma on the poten- 
tial can be determined according to the following relationship 

where X = v - Vq ; c — integration constant. In the case of v = Vq, dis- 
persion disappears, i.e., in the sense that Vq, is the limiting plasma 
velocity at which the electric field potential vanishes. We then have 

?=^[(T5C+l)-eT']. (9) 



146 



Since 



dv 1 



the electron plasma density may be found according to formula (9) 



» df 



J [U— V, (f)J' J [x— 1«>1" "A. 



n = «oef«. = n„e' 
Since 



J [X — 7?]^ J 1— e'" 1 — 



F(<p)=rto 






1— fiT' 



where xo is the value v = Vg at which the electron density equals the ion 
density. 

Consequently, the Poisson equation can be written as follows /153 

d^<P 2 1 



and X may be found from relationship (9) . 

Let us determine the characteristic of equation (6) for the linear 
dispersion law 

Let us assume that at an initial moment of time the field is switched on 
whose potential increases linearly with time: 

9o {'^) = To -^^ 
We then have , ~, \ t 

from which it follows that 



147 






Consequently, the characteristic Is real only up to a certain value x^^gy^^^j 
which depends on time 



bound ^^ L "PoTf J 



If ^>Xt^ound> *'^® signal which is switched on in the case of t = has still 
not reached the point under consideration. If x<X|jq^j^(J, the characteristic 

is determined in more than one way (it has two values) . The requisite 
value of the characteristic is determined by additional considerations. 



The propagation rate of the front boundary is 
Vbound=lP°""'=^^4i^+2M^. 



2 ~ AT 

Consequently, for y > the front velocity increases with the time, while 
in the case of y < it decreases. 

The velocity of a point with a constant given potential, as may be 
seen from (9), is determined by the wave phase velocity Vo((t)), i.e., by 
the specific value of the potential ^. We may formulate the potential and 
the electron velocity distribution function by the hydrodynamic parameters 
which are found. 



Electron Velocity Distribution Fimction in the /154 

Case of a Square Dispersion Law 

Let us find the electron velocity distribution function in the case of 
media with a square dispersion law 



The equation of characteristics 

dip 
dX 



g = -x,±i/=:2?. 



148 



where xi = u - Vq; u — electron kinetic velocity in the plasma, can be 
rated by substit 

We finally find 



integrated by substituting (j) = x?t 



u-\\,r \ -2y A -2y J 
^ 1/ ("-I'o 



X 



21/' -2? 

/ 5 — 1 
for a positive sign of the root, and 



--. I 



1 + 1^5 



(10) 



= Ci == const 



(u-V,)^\ _2y ^ /" -2y ] 

2*r L(u-Ko)^ y («-l'o)' 'J 

21/11=^ 

^ _ y (u- v^r 



X 



-, 1 



K5+1 



21/ -^y 



(11) 



= Co = const 



for a negative sign of the root. 

The solutions of (10) and (11) enable us to compile the distribution /155 
functions in the case of i}" "*■ which change into a Maxwell distribution. 
On the other hand, all the distributions changing into Maxwell distributions 
in the case of (J) ^0, in regions with a non-zero electric field, have the 
following form 

/ = y4 exp Ci for u > Vq, 
f = A exp C2 for u < V ^. 

Consequently, electrons have velocities lying outside of the boundaries 



«>Vo + 



2V-2<t 
/5— 1 



and 



7^5+1 



149 



The electrons whose velocities are included within 



" /5+1 ° /5— 1 

are damped by a wave, and are not included in the distribution functions 
of (10) and (11) . The relationships obtained enable us to study the 
collisionless transition of electrons from the region of trapped particles 
into other plasma electrons and the associated energy distribution. 



REFERENCES 

1. Akhiyezer, A. I., Lyubarskiy, G. Ya. , Faynberg, Ya. B, In the Book: 

Scientific Reports from Khar'kov State University (Uchenjrye Zapiski 
KhGU) , 6. Izdatel'stvo Khar'kovsk. Universiteta, Khar'kov, 73, 
1955. 

2. Faynberg, Ya. B., Nekrasov, F. M. , Kurilko, V. I. In the Book: Plasma 

Physics and Problems of Controlled Thermonuclear Synthesis (Fizika 
plazmy i problemy upravlyayemogo termoyadernogo sinteza) . Izdatel'stvo 
AN USSR, Kiev, 27, 1962. 

3. Nekrasov, F. M. Zhurnal Teoreticheskoy Fizlki, 33, 7, 1963. 

4. Khizhnyak, N. A. In the Book: Plasma Physics and Problems of Con- 

trolled Thermonuclear Synthesis (Fizika plazmy i problemy upravlyaye- 
mogo termoyadernogo sintez^, 1. Izdatel'stvo AN USSR, Kiev, 31, 1962. 

5. Smirnov, V. I. Course on Higher Mathematics (Kurs vysshey matematiki) . 

Vol. 4, Fizmatgiz, Moscow, 1956. 



150 



INDUCED SCATTERING OF LANGMUIR OSCILLATIONS IN A PLASMA 
LOCATED IN A STRONG MAGNETIC FIELD 

V. D. Shapiro, V. I. Shevchenko 

This article investigates the nonlinear interaction of harmonics in /156 
the long wave spectral region of Langmuir oscillations (kvDe ^"^ ^^ • '^^ 
linear damping of these oscillations, which is caused by the interaction 
with resonance particles, is negligibly small. It is assumed that the 
plasma is located in a rather strong magnetic field, so that the plasma 
particle oscillations are possible only in the direction of the magnetic 
field which is parallel to oz. 

The initial system of equations for the distribution functions of 
electrons and ions and the electric field has the following form 



5/1 ,,a V Sfl 



(1) 



X exp [—ilw^ ^ + u)_ — mj\ t); 

\ k-q q k ) 

k a J k \^J 

T i:k ^ -^ k 

k k k 

(fg is the background distribution function whose change with time can be 
disregarded, due to the small nimiber of resonance particles ( — -—j^Xd^ <^ l) 

and the small dissipation of oscillation energy during scattering). The 
notation in equations (1) - (3) is standard; summation in equation (2) is 
performed for plasma ions and electrons. We obtained the following rela- 
tionship for damping frequency and decrement, disregarding the nonlinear 
terms in the first equation, from equations (1) and (2): 



o_. = <o„,(l + I ^=x^^] cos e; T_. = I- . ^ cos 5^^ 



(4) 



CO— •■ 
k 



(e is the angle between the direction of oscillation propagation and the /157 
magnetic field) . The nonlinear interaction of harmonics leads to a change 
in the oscillation spectrum due to processes of wave decay and scattering 
by plasma particles. The laws of conservation must be fulfilled in the 



c. c. = complex conjugate. 

151 



case of two-plasma decays 

(u_» + (o_ = 0)^; ki+k2= k. (5) 

->• ->- 
Assiomlng that kj and k2 lie in one plane, we obtain the following condi- 
tion for the spectrum (4) from (5) : 

k-i cos G] it ^2 COS Bj 



Vk^i -f A| ± 2A,&2 cos (Gi — 02) 



= COS 9i + cos 62. (6) 



Decay is possible if k^^ ^ 0» ^22 < 0, which corresponds to the sign " — " 
in condition (6). In several cases, the spectrimi of Langmuir oscillations 
in a strong magnetic field is a nondecay spectrum, particularly if it is 
close to a one-dimensional spectrum O^ 2l 62*. 

Nonlinear wave scattering is caused by the interaction of plasma parti- 
cles with the beats of different frequency. This process becomes signifi- 

cant if the condition *'' '' 'f^ d„ is fulfilled. The beats which are 

the cause of wave scattering cannot arise due to decay, since the following 
condition is fulfilled for the waves formed during the decay 

(J) — (i)_^ <^_^ 
k[ ft. ft _ "0 -^ „ 

Therefore, in this case the processes of wave decay and scattering are in- 
dependent. The transformation of the oscillation spectrum due to the non- 
linear scattering process will be subsequently investigated. 

In solving the nonlinear equation (1) , we shall employ the method of /158 
perturbations. Substituting f", which is found from the linear theory, 

k 

in the nonlinear terms of this equation, we may employ (2) to obtain the 
following formulas for the electric field amplitude and the distribution 
function in the second approximation 



Ion oscillation branches are not examined. Actually, in the approxima- 
tion under consideration, decay of a Langmuir wave into a Langmuir 
ion-sound wave is possible. However, the energy primarily remains in 
high-frequency oscillations (Ref. 1). 



152 









X 






ft— 17 9 \ gy 



dv; 






f=-S 



k-9 



1 



X 



I ft-? 1 I 



X 



^4^ r i_ 






ft—? fl \ 9' -I 



dy. 



(7) 



In the third approximation, the system of equations for determining 
the electric field amplitude and the distribution function has the 
following form 



1- 

dt 



+ i {kzVz — U>- 



) r + -^£_, eft;, = _ _« ^ £11) ^ 
" k ' l^a kz '"a - \ ft-?. 






+ 



a/11' 



u)_. — a)_ 

q q kj 






(8) 



f2) (2) 
where E'-.,. , f ^ are determined from formulas (7). We obtain the kinetic /159 

k k 

equation for waves from these equations by simple computations: 



-^ = i^E'^ - / 2 // (k, 9. x\ E E X 

"' k kz -— \. I k-qz q — xz 

q,x 

X eXp f i /"ffl-, _ + 0)_^ ^ + 0)^ iO^\ t\. 



(9) 



153 



where y"^ is the decrement of Landau linear damping: 
k 



// {k, q, x) 



k—q q — X X fe 

\ ft— 9 q—7. X J 



- X 



X ( 






X 



k — q q — X X 






d I 3d, 



?^^z 



^^z \ '-z^z — «■-. , 






X 



17 — X X 

.3 



dy + 



---rV 



\ <7— X X j 



;9^^'"' 



4< p 1 

J *— ? q— 



q — X X 



dv. 



I 1 \ dfl 

Qz"z —"> ► — "-. (*z — 9z) "j — »-. ^1 3^2 

?— X X k~q I 



X 



X 



4iteS 



d I 3"' 



a '"p 



m^ J 92^2 — "^ - — ' 



dv^ \ x^u^ — '"- 



\dv ) 



9— X 



Multiplying (9) by E , averaging over time (only the terms with 



kz 



-^ ->■ 



/160 



X = k and x = q - k, which change slowly with time, remain in the right- 
hand side) and combining the equation obtained with the complex conjugate 
one, we can write 



dt 



kz\ k\ kz] -^k q \ kz\ \ q \ 



(11) 



.Y^ = ^Im ( H{k,k~q,q)+H{k.k—q,k) ) 



1 V 4tei 



dfo 
dv. 



k 



('■'■-i)' 



X 



X 



^^ + fc^x 



{,z^z-^^)[{K-^z)^z-^^+<^^\ \X_-;\ 



154 



X ^^rx 



V ft 9 ) 






dv] ). (12) 



In computing the Integrals Included in equation (12) , we assume that 
the following conditions are fulfilled: o) >> kv^, which corresponds to 

k 
weak wave absorption, and f^^— w » which is necessary in order that a sig- 

k q 
nificant number of plasma particles may participate in the interaction 
with beats of different frequency. Let us investigate the case when the 
contribution made by Ions to the plasma polarizability at a different 

frequency is negligibly small — i.e., the condition , __ r^- 'v 

'^ m bn* C 1- is fulfilled. Then, assuming that the distribution func- /^^^ 

"^i De 

tion of the plasma particles fg is a Maxwell distribution, we obtain the 
following from equation (12) after very cumbersome computations 

T" (k, 6. <7. e') = - ? l/l • ;;4-cos 6F,^ f^>^.. X 

/, , 2 ( cos e - COS 9' ) \ ( m <4jcos^j-cos£)^| 

^ r + ¥ ■ (k-^- q^) l-'^, cos 9] ^^P l~ 2T, ' (k - qr cos^ 6 f 

The increment y \k,e,q,0') which determines the Induced scattering rate 
of plasma oscillations in k-space differs considerably from zero only in 

the narrow range of angles: |8' — 6|:S — in the case of 6 t= and 



1 9 _ 6' I < t/ ^ in the case of 9 :^ . 



Thus, during scattering the spectrum which is initially one-dimensional 
remains close to a one-dimensional spectrvim. In the case of 9' - *-*-, it 

coincides with the increment obtained previously in a one-dimensional model, 
within an accuracy of the factor cos (Ref. 2, 3). It increases somewhat 

with an increase in 9 - 9' ^*'^~ up to f'^a^^yi' W+W^X^^ '^^'^' ^^^ 

then rapidly decreases to 0. 



155 



illllllil 



In the real case of a confined plasma the spectrum of the values 

Vrj. 

is always discreet, and for a sufficiently small ratio — =^, only one 

v$ 

possible value 6' = 9 can enter the angular range 0' - in which y 
is large. The scattering of plasma oscillations in a strong magnetic 
field then takes place, in fact, in the same way as in a one-dimensional 
model, in contrast to scattering of oscillations in a plasma which is not 
located in a magnetic field. In the latter case, scattering at large 
angles is possible, and, if the angle between the wave vectors of two 

interacting waves considerably exceeds , y^^i^icreases by a factor of 

1 Q Q 

as compared with y in a one-dimensional model (Ref. 4, 5). The 



k2x2 

De 
contribution made by ions to the plasma dielectric constant is quite sig- 

("0/ cos'' 9 
nificant when the condition ~, n- ~:^ 1 is fulfilled. We should note 

I ^ 7) 

that this condition is fulfilled most readily in the case of 2i S, when 

the difference to - o) is at a minimum: m., — u)_ =^ -h- u'o^ cos ()fe* — 9^)^ J ^^'^ /162 
k q * , ^ 

wq,- cos' 9 m, 1 /«,, 1 m, 

K~^^«irf ^ «('fe^* Assuming that the conditions TR}' ^>^ ^ m^^ 

X . <g 1, are fulfilled, we obtain the following expression from (12) in 
the case of 6 = 0' for y^^ 

T"(/^.0.,.6) = -^y ^. ^^-j ,-^_^. _-_^_. (14) 
i.e., in this case y increases somewhat as compared with (13). In the 



2 



9' /cos G _ cos 8' ~ :il\ ^ -'' - 



m. 



case of G =5t 0' /cos — cos 8' —\ ^ x 1 and when the condi- 






tion — 



«<. 1 



(13). 



■C 1 is fulfilled, y is determined in this region by formula 



We should recall that this investigation pertains to a case when the 
plasma is located in a magnetic field which is so strong that the plasma 
particle oscillations are only possible in the direction of the magnetic 
field. The condition of "magnetization" of the electron component, as is 
customary, has the form coog << Wjje* i.e., it is fulfilled for field 
strengths which are not too large. In this case, when the ions make a 



156 



significant contribution to the dielectric constant of the plasma, it 
must be required that the ion plasma component be "magnetized". The 
corresponding condition is harder: 






— <0_» I < U)//,a . e ; <"//« > Woe — ' ^^^oe- 
q I '"« 



Let us turn to certain general characteristics of the nonlinear 
change in the wave spectrum. Since the Increment of nonlinear scattering 
7^ (k,e,q,e') in the one-dimensional case 6=0' changes sign when k^ q 
is substituted [see formulas (13) , (14) ] , the change in the total oscilla- 
tion energy during scattering in the approximation under consideration 
equals zero 

fc k'q ^ -^' 

Employing formulas (13), (lA) , we may also readily see that the non- /163 
linear interaction of harmonics leads to a transfer of energy along the 
spectrum to smaller k: 



S^l^-r^ Y.\^^-9UHk,^,,,n\E^\Y-;\' 



(16) 

* k q 

The total oscillation energy during scattering by plasma particles 

changes in the subsequent series with respect to k^X'^ , since in this 

De 

approximation an addition to y^ appears , which is symmetrical with respect 

to the k 4?" q substitution. Assuming, for purposes of simplicity, that 
the oscillation spectrum is one-dimensional, we obtain the equation for 
the change in the total energy in the spectrum: 



avlr:.!*. V fn T. "''^'D' "^ 



1 3 



k-\%. 



k 

(17) 



The second term in this equation which describes the oscillation energy 
change during scattering becomes more significant than the first term, 
which characterizes the oscillation energy change as a result of their 
interaction with resonance particles, when the following condition is 
fulfilled 



157 





3 ' 


1 — 2 « 2*^X2 


oO ^,0^10 



(18) 

i.e., for relatively small oscillation amplitudes |e, p « NqT, if the 
parameter is fairly small. 

The possible dissipation of oscillation energy when they undergo 
nonlinear scattering by plasma particles was pointed out in (Ref. 6, 7). 
However, for Langmuir oscillations this phenomenon is k^X^ times less 

than the change in the field intensity in separate harmonics in the spec- 
trimi due to energy transfer. 



REFERENCES /164 

1. Orayevskiy, V. N., Sagdeyev, R. Z. Zhurnal Teoretlcheskoy Fiziki, 

32, 1291, 1963. 

2. Drummond, W. E. , Pines, D. Yademyy Sintez. Appendix, 3, 1049, 1962. 

3. Shapiro, V. D. Authors 's Abstract of Candidate's Dissertation 

(Avtoreferat Kand. Diss.). Ob ' yedinennjry Institut Yademykh Issle- 
dovaniy, 1963; Zhurnal Eksperimental'noy i Teoretlcheskoy Fiziki, 
44, 613, 1963. 

4. Gorbunov, L. M. , Silin, V. P. Zhurnal Eksperimental'noy i Teoretl- 

cheskoy Fiziki, 47, 200, 1964. 

5. Gaylitis, A. K. , Tsytovich, V. N. Zhurnal Eksperimental'noy i 

Teoretlcheskoy Fiziki, 47, 1468, 1964. 

6. Kadomtsev, B. B., Petviashvili, V. I. Zhurnal Eksperimental'noy i 

Teoretlcheskoy Fiziki, 43, 2234, 1962. 

7. Karpman, V. I. Doklady Akademii Nauk SSSR, 152, 587, 1963. 



158 



NONLINEAR THEORY OF LOW FREQUENCY OSCILLATIONS EXCITED 
BY AN ION BUNDLE IN A PLASMA 

D. G. Lominadze, V. I. Shevchenko 

As is well known, in the case of the interaction between a bundle 
of rapid electrons and a plasma, the energy lost by the btmdle during 
relaxation changes into thermal energy of electrons in the plasma and in 
the bundle, and into energy of high frequency Langmuir oscillations 
(Ref. 1). When investigating the possibility of heating the plasma ion 
component during bunched instabilities, it is of interest to investigate 
the excitation of low frequency oscillations by the bundle, in the non- 
linear approximation. A previous article by D. G. Lominadze (in collabora- 
tion with K. N. Stepanov) investigated the linear theory of low frequency 
oscillation excitation in a plasma located in a magnetic field by an ion 
bundle. In a strongly non-isothermic plasma (T^ >> T^^) , during the passage 

of an ion bundle, longitudinal longwave/ 4- J_a << l\ oscillations may be 
excited, whose frequency is determined by the following relationship 

'l = ^ (">' + <d ± y [("^s + '^hy-^«i ^°^' H '''' (1) 



U) 



Wl^e^e »?. Un„e- . ^2 _ T. 



1 + 



M ' De \T^nJ^ ' 






Wjj^ g is the Larmor frequency of ions and electrons, respectively; — /165 
the angle between Hg and the direction of oscillation propagation. If 
T^ "^ Tg, the ion bundle can excite longitudinal shortwave oscillations 



f-e^^) 



in the plasma with the frequencies (Ref. 2) 



U)„ = nu)w,(l + tpn); '^n = 



(-9 

tt = 1, 2, . . 



e h -^rz^ (2) 



k|yxi 
where y^^ = ; Ijj(yj) is the Bessel function of the imaginary argument. 



0)2 
Hi 



The process by which bunched instabilities develop may be 



159 



dynamic) stage, the bundle remains monoenergetic ( « 1, v, 

k 



qualitatively divided into two stages (Ref. 1). In the initial (hydro- 

Tl ~ 

thermal velocity of bundle ionsl and instability develops veiry rapidly; 

j^j '^(/Iq. "i — densities of plasma and bundle, nx « ng, co — 
excitable frequency). For rather large oscillation amplitudes, the 
thermal energy in the bundle is so large I rr~ — 1 1, that the bundle 



relaxation may be investigated in the quasilinear approximation (quasi- 
linear stage) . The time required for the development of this instability 

T^O 1 

TniiaGT is on the order of — • — . 
quasx j^^ (jj 

This article investigates the development of instability at the first 
stage, the most unstable oscillation branches are found which produce the 
dynamics of the instability development, and the change in the macroscopic 
parameters of the bundle and the plasma is determined (thermal energy, di- 
rected velocity) in the case of instability. In a strong magnetic field 
("^i ^^ '^s^ it is possible to trace the development of instability at the 
quasilinear stage and to determine the state at which the bundle and the 
plasma arrive as a result of the quasilinear relaxation process. 

Excitation of Longwave, Low Frequency Oscillations /166 

(Hydrodynamic Stage) 

We shall assume that at this stage the following condition is ful- 
filled 

-e«^'-t-«' (3) 

k k 

(6u is the change in the bundle velocity at the initial stage) , at which 
the dispersion equation of longwave oscillations has the form 



"01 



2-sin2 9=0. 



When investigating the dispersion equation (4), we can examine two cases. 

160 



1. a)jj£ « 0) (weak magnetic field). During Cherenkov excitation 
(o) = k^ Uq), oscillations with the frequency 



(,+l.|3i„.»). 



have the largest increasing amplitude increment. The maximum value of 
the corresponding increment is 

l^(^V\>.rn,V.9. (5) 






The formation of instabilities in this case is possible in the case of 

cos 6 < — ^, if Un > v„ , and for any cos 6 , if u„ < v„ I v„ =^/_§. 1. 
Up OS ° ^ V Tm / 



2. a)jj£ >> (iOg (strong magnetic field). In this case, ion-sound waves 
the frequency to = w cc 
increment of these waves is 



with the frequency to = w cos 6 have the largest increment. The increasing 



V 



(6) 



Excitation of oscillations is possible if — H. > 1. 

"0 

Relationships (4) - (6) were obtained by disregarding damping by /167 
plasma electrons, which is valid if the following conditions are fulfilled 






If condition (3) is fulfilled, all of the bundle particles are in resonance 
with the wave, and its hydrodynamic description is possible by means of 
the moments of the velocity distribution function 



^ = 1^ v^dv; n=i-J {^' - "^) {"" - "*) ^S^''- 



We can obtain the equations describing the change in these quantities with 
time from the kinetic equation of the Fokker Planck type, which is derived 
in (Ref. 3) 

at -dviY-ii'dv^)' (7) 



161 



3 = e, i pertain, respectively, to plasma electrons and ions and bundle 
ions; a^j^ — diffusion coefficients in velocity space: 

e^ V r p 12 *^^ V" "°'^" ^^'^^ rr 4- T V 



n=l 






(8) 



n=l 






'_„ + r„]. 



Here we have 



"±n — TT 






/(o':^-;i^u^± «<»„,.y^ 



The frequency (ji)-> and the increment Y"^ ^'^^ determined by the disper- /168 
K k 

sion relationship of the linear theory (4) . 

We obtain the system of equations for the change in the "bundle" 
parameters due to the development of instability from formula (7) 



^ _ 2^2 y If IM ^ ^ 

dt AT- ^ I k\ \ k^ 



('-"•- "r) 



[(v.-.y'H-e] 



~ fc2 /I... -^ 



^^ k'^ 4a. 



«i 



X 



7-* 
k 



7-. 

* 



L^ ft ) k \ k Ik. 



dt 



* ^ *'' ft 



X 



+ ■ 



(*."o - -1 - »«,)'+ 7l (*,«o - <»: - ««,)'+ d 



(9) 



162 



The change with time of |e | ^ is described by the following equation 

I * „. ,^ ,2 (10) 



(it 



k I k 



We may obtain the equation for the change in the plasma ion para- 
meters from formulas (9), assuming that u^ = 0. The condition for the 

kAv^ 
applicability of a hydrodynamic description of plasma ions — << 1 is 

less hard in the case of n^ « ng than equation (3) , and this description 
is applicable throughout the entire development of instability. 

Integrating equations (9) with respect to t, we obtain the equations 
determining the change in the energy of directed bundle motion and in the 
thermal energy of ions of the bundle and the plasma. 



In the case of a weak magnetic field (wj,. << Wg) , we have /169 

ni/wwoSw = - 4^ • ^Sp^l'; 

k ft 

In the opposite case (Wrj. >> Wg) , we have 



o^„,W \ t. ^^7«K^ . ft. "C>^-,, A, (-Li) 

„2 






'-7 • 

163 



The change in the electron distribution ftmction in the case of 
instability is determined according to the following equation 



<}fo_ dV 1 r^tiP [2:^.__^iL 1° 



4^J^*l^rr*^"(<oj-*Af+7i 



dt - dvAm'- • (27:)' I "" r r "' fcoi _ A o \2 + ^l dv 



(13) 



The phase velocities of the excited oscillations change between Vrj.^ << 
<< Vj, << Vrp . Therefore, the main portion of plasma electrons is in /170 
resonance with the waves v >> v^. The change in the electron distribu- 
tion function for these v is described by the equation 

[ k } 

Thus, for a change in the electron thermal energy in the case of insta- 
bility, we obtain the following in the case of a weak magnetic field 



and in the case of a strong magnetic field 



«s 



I.„.,7-„=_^.l2|£.p. (16) 



1- 2 



For plasma electrons which are in resonance with excited oscillations, 

whose velocities lie in the narrow range — ^ «/^ , we obtain the following 

■^Te » " 

expression from equation (13) 






|-.iwf''»rrr5M"r-*-"-)'^"|- <!'> 



Changing to the variable $ =-^( I / ~<i<l] in formula (17) and 
representing fp in the following form 



164 



f^='U-T^),.(^A (18) 

where (p (?, ^) 1,^0 = 2"' ft\v±] ^^ ^^^ portion of the electron distribution 



function depending on vj_, we obtain the following equation for the change 
in (t)(C, t) 

dt In. ' m^ £,5 



{lVrrS'"T.s}- (19) 



The Interaction of resonance electrons with oscillations leads to /171 
the occurrence of a plateau in the electron distribution function in the 
region of excited oscillation phase velocities. The time required to 
establish the plateau may be determined from formula (19) : 

* 
(A? — the dimensionless width of the plateau). 

The change in the energy of resonance electrons is 



Mj \/n / ft ft 



(21) 



i.e., it is considerably less than the change in the total energy of plasma 
electrons. 

We may determine the balance of energy during the development of 
instability from formulas (11) and (15) , (12) and (16) : The energy of 
directed motion, which is lost by a bundle, changes into the thermal 
energy of particles in the bundle and the plasma and into energy of 
electrostatic oscillations. Since conditions (3) must be fulfilled in 
the initial (hydrodynamic) stage, we may readily estimate the maximum 
energy of oscillations excited at this stage: 

r 

At this oscillation energy, the time required to establish a plateau in 
the electron distribution function, as follows from formula (20), is 



165 






(23) 



i.e., it is a little less than the time required for the development of in- 
stability in the initial, quasilinear stages. 

Thus, the plasma electron distribution function changes most rapidly 
in the resonance region: A plateau appears in this region in the velocity 
range in which oscillations are excited at a given moment of time. A 
change in the spectral density of oscillations and ion parameters occurs 
much more slowly. Under these conditions, the electrons have no signifi- 
cant influence on the dynamics of the instability development. 



Excitation of Low Frequency Oscillations^ 
(Quasilinear Stage) 



/172 



Further development of instability leads to a still greater increase 
in the thermal scatter in the bundle, and its distribution function be- 
comes so diffused that the quasilinear approximation is applicable. 

In the case of to >> ojg^, it follows from the expressions for the 
diffusion coefficients (8) that aj_j_ '\' otzz* ^^ ^^"^ '^ '^' '^^ problem is 
thus essentially a three-dimensional problem. Let us investigate the 
opposite case, a)jj- » tOg, since in this case a_Lj_ << a^z at the quasi- 
linear stage, i.e., only longitudinal diffusion is significant. 

The initial system of equations for the quasilinear approximation 
has the following form 



dt " Ov 



^'?Krr>(";- 



I k I 
dt 



-*.".) g 



r «! '"sdg 



* 



(24) 



(25) 



Here g(t, v) is the distribution function of bundle ions integrated with 
respect to vj_. 

Substituting r— °- Ie p from equation (25) in formula (24), changing 

k 



166 



from summation over k to Integration, and integrating with respect to k 
due to the 6-f unction, we obtain 



"01 






1a1\ 

dt 



(26) 



Integrating with respect to v and t, we may write the following 
equation 



71/2 

I sin 0d6 f I £_, (/) p _ i £_^ (Q) pi = 2-k 

11*1 \ k I / 



X U^ 1 — 



_9^2^I. "1 J_x 



^ n^(^ j'')-^(o, y')]d^'. 



2\Vl O 



(27) 



where vj is the lower boundary of the instability region caused by damping /173 
by plasma ions. We may employ this equation to determine the oscillation 
energy at the quasillnear stage: 



W = 



i^|£.(oo)p=i4^„,j;|^i_ljji^ 



(oo, V')- 



(28) 



-g(0, v')]dv'dv. 



where g(«>, v) is the distribution function at the end of the quasilinear 
stage — the plateau, whose height is determined from the following condi- 
tion 



^(oo, v)ivo — Vi) = ni; g{co,v) 



(29) 



v^ and V2 (the upper boundary of the instability region) are determined by 
the following relationships 



^(oo, fi) = g-,- (yj); g(co, v.)=go(v2y, 



(30) 



go(v) — the distribution function of the ion bundle at the beginning of 
the quasilinear stage; g.(v) — the distribution function of plasma ions 
integrated with respect to vj_. We obtain the following from (30) 



/2T- 



/Jn"n 



2r.T,\'/' 



-m 



««o; t'.-«o(i-(^J'). (31) 



Performing integration with respect to v in (28) , we may write 



167 



The change in the energy of the plasma ion thermal motion may be deter- 
mined from (11) : 

i^o^Tu = -hl^\E^l4-^=i^''^"l- (33) 

I r 

The increase in the energy of the plasma electron thermal motion is 



k 

The finite bundle velocity established at the end of the quasilinear stage /174 
is 

■ Vt 



""= J Vg(ca, v)dv^-^^^^^. 



Vl 



i.e., the energy loss of the ordered bundle motion 



8e=-4rtiM«2. (35) 

O 



_3 
8 

The thermal bundle energy acquired during the development of instability 
is 

n,hT, 



- = J^(~. v)[v l^j _d„^__. (3gj 



We may employ formulas (32) - (36) to verify the fact that the law of con- 
servation of energy is fulfilled 

k 

Thus, in contrast to the excitation of high frequency oscillations, 
the excitation of low frequency oscillations leads to the transfer of a 



(4)o. 



considerable portion I'^r-jof the bundle energy to plasma ions, and also 
leads to significant heating of the ion component. 

The quasilinear theory disregards the nonlinear phenomena of oscilla- 
tion scattering by plasma particles. As was shown in (Ref. 1), in the case 



168 



of the excitation of Langmuir oscillations by an electron bundle, non- 
linear interaction of harmonics is usually insignificant in the quasi- 
linear stage. However, there is a parameter region in which these phenomena 
determined the dynamics of the instability development (Ref. 4). 

We shall continue to study the role of the nonlinear interaction of 
harmonics in the case (which we are considering) of excitation of ion- 
sound oscillations by an ion bundle. 

Excitation of Shortwave, Low Frequency Oscillations /175 

(Hydrodynamic Stage) 

When allowance is made for the finite, Larmor radius of plasma ions, 
excitation of longitudinal oscillations by the harmonics '^^<*Ht. is possible. 

One important feature of these oscillations is excitation in the case of 
T^ '^ Tg and propagation almost perpendicularly to the magnetic field. 

Therefore, it is more likely that low frequency oscillations are the 
reason for anomalous plasma diffusion perpendicularly to the magnetic 

field, rather than ion-sound oscillations, for which — << 1. 

K 

Let us investigate the manner in which the ion bundle and the plasma 
parameters change at the initial excitation stage of low frequency oscilla- 
tions. 

The dispersion equation for shortwave oscillations for conditions (3) 
has the following form 

Ti V ""'"(C) ""oi 2fi "01 y 



n= — oo 



X Sin^ 6 == 0. 



Waves with frequencies of Wjj = nojg^ (1 + ^^ have the largest increment. 
Their increasing increment in the case of n = 1 is 



where 



- 2 \n, 
a{k) 



/l(^)' 



169 



The change in the ordered velocity and thermal energy of bundle ions 
may be described by means of (9) , where w and y 3.re determined from 

k k 
formulas (38) - (39) . Making use of the fact that y^ has a maximimi with 

respect to k, we may obtain the following in the case of instability at 
the frequency co = cojjj (1 + ^i) 



M / *=*, 






(40) 






' ' ' <"?/.■ L. I'- 1"» ^^«oV ."^ 



^2 -V 0/ a /.(ft) 

J ft=ft. 

(kg — the value of k at which y has a maximtrai) • Let us employ the /176 

k -^ 

approximate formulas given in (Ref. 2), and we shall assimie that a(k) 

reaches a maximum in the case of -^° ^^ ^i^ } 5; 

a(I)^_^^«l; iil^£L<. (41) 

Assuming that << )|)i « 1, we obtain the following expression for the 

plasma ion diffusion coefficients in the case of instability at the fre- 
quency (1) = a)jj^ (1 + ii)i) 






a -^. -l-{dk\F |2^^ " ' """' 



e- 






n=I 



7-. 
X - ^— 



(42) 



' + ll 



170 



II I II 






n=l 
X_^ 
k 
X 



/?(xi) 

2 



,. X 



We obtain the following by means of (42) from formula (7) : /177. 







k- d^i^Yrf 



(43) 



M 
We may determine the change in the plasma electron energy by formula (14) 



1 „^8r . . = /'4;^ 8^ • (ripj ^^1 ^r r * ^''''^ 



k" — /- - 

Comparing expressions (40) , (43) and (44) , we can see that the 
energy of the bundle ordered motion changes primarily into energy of the 
transverse thermal motion of plasma ions, and may lead to significant 
ion diffusion perpendicularly to the magnetic field. Relationships (40), 
(43) and (44) are valid as long as the conditions of a monoenergetic bundle 
are fulfilled (3). 



The maximum energy of low frequency fields, obtained at this stage. 



IS 






REFERENCES 

1. Shapiro, V. D. Zhumal Eksperimental'noy i Teoreticheskoy Fiziki, 33, 

613, 1963. 

2. Drummond, W. E. , Rosenbluth, M. N. Phys. Fl. , 5, 1507, 1962. 



171 



3. Shapiro, V. D. , Shevchenko, V. I. Zhurnal Eksperimental'noy i 

Teoreticheskoy Fizlki, 42, 1515, 1962. 

4. Tsytovich, V. N. , Shapiro, V. D. Yademyy Sintez, 5, 1, 1965. 



NONLINEAR PHENOMENA IN A PLASMA WAVE GUIDE 

(ION CYCLOTRON RESONANCE AT A DIFFERENCE 

FREQUENCY) 

B . I . Ivanov 

Ion cyclotron resonance (ICR) has been extensively studied both /178 
theoretically and experimentally [see, for example, the summary in 
(Ref. 1)]. Several works have appeared recently (Ref. 2-4) which 
examined the problems of the nonlinear theory of ion and electron cyclo- 
tron resonance. As is known, the non-linearity criterion has the 
following form (Ref. 5, 6) 

eEpl /. __^\ — ' , 

V$ 
(Eq — strength of the wave field; X — wave length; 3^ — retardation; 

vq — ordered plasma velocity) . Formation of nonlinear phenomena is 
facilitated during resonance (Ref. 6) , since in this case the non-linearity 
parameter contains the additional factor "^ w (oj-to )~^. Thus, it is 
possible that nonlinear phenomena may occur in the case of ICR, because 
for this case the occurrence of large strengths of the wave field, small 
phase velocities, and large wavelengths is characteristic. In this case, 
nonlinear phenomena may play a significant role during heating (nonlinear 
damping, nonlinear shift of the resonance frequency) and during the intro- 
duction of high frequency energy into the plasma (interaction of frequencies) . 

In principle, it is possible to introduce large UHF power into the 
plasma at two frequencies , and then to perform ICR at the difference fre- 
quency. Such a mechanism is also possible during the excitation of low 
frequency oscillations in the plasma-bundle system (Ref. 7). As is known, 
in unstable plasmas, low frequency oscillations, whose origin is 

172 



sometimes difficult to establish, occur simultaneously with high fre- 
quency oscillations. The occurrence of the high frequency oscillations 
is satisfactorily explained by the theory of plasma-bundle interaction. 

One of the mechanisms leading to the formation of low frequency 
oscillations may be the nonlinear interaction of high frequency oscilla- 
tions [separation of difference frequencies, decay instabilities (Ref. 8, 
9)] with the subsequent transfer of energy from high frequency oscilla- 
tions to low frequency oscillations [parametric amplification (Ref. 10)]. 
This article makes an attempt to provide a model for this mechanism by /179 
which low frequency oscillations are excited close to the ion cyclotron 
frequency. The parameters of the apparatus are the same as in the pre- 
ceding article (Ref. 12), which investigated the nonlinear distortions 
of the signal form and the formation of combined frequencies . We employed 
generators having a small power ('\'l w) which had a relatively small dis- 
turbing influence on the plasma which was produced independently. In 
order to fulfill the non-linearity condition, it was necessary to operate 
at low frequencies (f '\' 1 Mc) and with low phase velocities (3$ '^ 10~^) , 
which, in its turn, made it necessary to employ a low-density plasma 
(n 0- 10^ cm-3) (Ref. 12). On the other hand, f_ ^ f^^ >> v^^ (\)^^ "- 10^ p 

— collision frequency) represents the necessary condition for observing 
the ICR at the difference frequency. In view of these considerations, 
the main ("beat") frequencies and the difference frequency were of 
order of magnitude one: f i '^ f2 '^ f_ '^ 1 Mc. 

In order to observe the weak ICR signal, a sensitive system of a 
balanced, high frequency bridge was employed (measures were taken to re- 
duce the noise level) . Figure 1 shows the diagram of the apparatus (1 — /180 
current regulator; 2 — voltage regulator; 3 - high frequency generators; 
4 — amplifiers; 5 — phase inverter; 6 — phase rotators; 7 — AVC unit; 
8 — two-ray oscillograph; 9 — heterodyne receiver; 10 — heterodyne fre- 
quency meter; 11 — self-excited oscillator; 12 — main anode; 13 — quartz 
tube; 14 — water cone; 15 — auxilliary anode; 16 — cathode; 17 — mag- 
netic field recorder; 18 — palladium filters). The plasma wave guide 
consisted of the following, which were distributed coaxially: A plasma 
core with a diameter of 1 cm, a lead tube with a diameter of 3 cm, a 
copper casing with a diameter of 23 cm with the cross section along the 
generatrix, and a solenoid. The total length was about 180 cm. The 
quartz discharge tube was evacuated from both sides to a vacuum of 
'^ 10~ n/m^, after which hydrogen was introduced from both sides through 
the palladitm filters. When the entire length of the tube was continuously 
evacuated, it was possible to obtain a constant pressure. Before the 
measurements, the tube was treated to preliminary processing with pro- 
longed, high frequency discharge (wavelength X = 2 m, generator power 
P '^ 500 w) . The plasma was produced by discharge at a constant current, 
and the discharge current was stabilized. In order to increase the 



173 




to pump 



GDHZD 



3 [^ 



10 



ff 



to pump 



Figure 1 



jnization coefficient, a cathode was employed made of lanthanimi hexa- 
Dride with indirect direct current heating . Two anodes were em- 
Loyed in order to obtain a stable discharge: The main anode and the 
iixilliary anode. The anode potentials were selected according to the 
Lasma noise minimimi. 



The main frequencies fi and f2 from the generators were excited in 
ae plasma wave guide by short spirals located close to the left end of 
tie wave guide. An adiabatlc, absorbing water charge (length of about 
cm) was located at the right end of the wave guide. The reflection 
oefficient from the right end of the wave guide k equalled 0.1 - 0.3 
Ref. 13). Thus, moving waves with main (fi and £2) and combined 
nfi +mf2; n and m — whole numbers) frequencies could be propagated 
a the wave guide. The output signal was employed on two spirals, one 
f which was located in the magnetic field section which could be 
odulated by a commercial frequency. From the receiving spirals, the 
ignal was supplied to the two arms of the high frequency bridge. Each 
rm consisted of an amplifier, a phase rotator, and the AVC unit. The 
atter was used to eliminate relatively slow unbalancing of the bridge 
""^ Avr ^^ ''■mod^ ' "^^ signals from the arm of the bridge were supplied 
ut of phase to the heterodyne receiver adjusted to the resonance fre- 
uency, and after rectification through the low-frequency filter they 
ere supplied to the oscillograph. In the normal position, both arms 
ere almost completely balanced, but the amplitude of the signal with a 
odulating frequency remained somewhat larger. During the modulation 



/181 



174 




Figure 2 



of the magnetic field, at the moment the resonance value was crossed 
(f_ — f„j) — twice during the modulation period — a signal for the 

bridge unbalance was produced, which could be observed on the oscillo- 
graph. This recording system has good sensitivity and noiseproof qualities 
(subtraction of the signals from the two receiving channels leads to an in- 
crease in the modulation depth of the carrier frequency by the ICR signal, 
and simultaneously eliminates the plasma noise correlations along the wave 



guide length) , 

Figure 2 presents oscillograms showing the dependence of the reso- 
nance position on the magnetic field strength (fj^ = 1.5 Mc, f2 = 2.5 Mc, 
f_ = 1.0 Mc; n = I-IO^ cm-^, p = 7-10-2 ^/^^ ^ ^a = ^ ka/m, H 1 56-62 ka/m) . 
The upper line corresponds to the resonance signal which is inverted 
during rectification. The lower line corresponds to the signal coming 
from the generator recorder of the magnetic field. With an increase in /182 
the constant magnetic field strength (for a fixed difference frequency 
and a constant amplitude of the variable magnetic field) , the resonances 
converge, since the resonance condition (f_ 1l Jci^ -^^ fulfilled in the 
negative half period of the variable magnetic field. 



Figure 3 shows the dependence of the resonance position on plasma 



'\j 



density (fi = 1.5 Mc, fa = 2.5 Mc, f_ = 1.0 Mc, H = 62 ka/m, H^^ = 7 ka/: 



m. 



175 




Figure 3 



p = 7-10-^ n/m% n :^ (1 - 2) 10^ cm 
density, the resonances diverge, i.e 
large magnetic field strengths. 



^) . With an increase in the plasma 
, they shift into the region of 



In order to obtain quantitative estimates, the reflection coeffi- 
cients from the wave guide ends, the magnetic field strength, the phase 
velocity, and the plasma density were measured. 

Dynamic measurements of the magnetic field strength were performed 
by the generator recorder (Ref. 11) (see Figure 1). The self-excited 
oscillator circuit was located in the modulated section of the solenoid. 
Carbonyl iron was used as the induction core. Due to the small dimen- 
sions and the small value of y ('^ 10) , the recorder disturbed the magnetic 
field to an insignificant extent. With a change in the magnetic field /183 
strength, due to the dependence y(H) the circuit inductance and the self- 
excited oscillator frequency changed. The latter was measured by the 
heterod3me frequency meter according to the zero beats , which could be 
recorded simultaneously with the ICR signal by the two-ray oscillograph 
(see Figure 2) . This system was calibrated initially by nuclear magnetic 
resonance. 

The phase velocity was measured by the system shown in Figure 4 
(1 — Ave unit; 2 — phase meter; 3 — amplifier; 4 — delay line; 



176 



^^^^m^^i^^^^^^^^ 







.,^_ 


fTT=>r- 




T-K^f-- 














r-^ 


^ 


^^ 


^ 


-- 


! — -\^^ — -jj^sy^-^" — — '5^::=ii;r7-" 






K6?s5«w9©g9«^5^»oo<<)<»<^>^^ 
































f 




2 




3 


^ 


4 




5 


— 










1 












































. L 






















/ 


J 




1 






J 









Figure 4 



5 — limiting attenuator; 6 — generator; 7 — selective microvoltmeter) . 
A wave with the frequency f = f_ 2l f d was excited In the plasma wave 
guide by the generator. A high frequency signal was applied to the two 
antennae (a non-mobile short spiral and a mobile whip), and was supplied 
to the phase meter input. The compensation method was employed to 
measure 3$. A coaxial delay line, consisting of segments of the high 
frequency cable (length, 2.5; 5; 10 and 20 m) , was switched into the 
circuit of the mobile collapsible-whip antenna. These high frequency 
cable segments could be subsequently combined in any combinations. With 
a minimum distance between the antennae, the delay line was completely 
introduced, and the phase meter indicator pointed to zero. As the dis- 
tance between the antennae increased, the delay line decreased so as to 
compensate for the phase shift produced. In spite of the fact that the 
phase meter system was designed so that the phase reading was not depen- 
dent on the signal amplitude, there was a possibility of error for small 
phase shifts and significant changes in the signal amplitude. In order 
to eliminate this possibility, an AVC unit was switched into the circuit 
of both antennae, and also a selective microvoltmeter (for controlling 
the high frequency signal amplitude) and a limiting attenuator (for main- 
taining the amplitude at a definite level) were introduced into the cir- 
cuit of the mobile whip. 



/184 



The length of the delay line AL changed linearly as a function of 
the distance between the antennae £. The retardation was determined 
according to the following relationship 



ALV 



=- ~3. 10-2 



177 



(e — dielectric constant of cable insulation) . 

The electron density n was measured by the shift in the eigen 
resonator frequency Af , and the dependence Af (n) was determined by pre- 
liminary calibration with respect to the electron bundle (Ref. 12). The 
relative content of atomic and molecular hydrogen ions in the plasma was 
not measured. 

The following conclusions may be drawn on the basis of the measure- 
ments performed. At the moment that resonance is passed, the amplitude 
of the "difference" wave increased. The resonance frequency increased 
with an increase in the magnetic field strength, and decreased with an 
increase in the plasma density, while f_ — fci« The picture observed 
corresponds qualitatively to the excitation of ion cyclotron waves at 
the difference frequency. 

The experimental data agree qualitatively with the dispersion rela- 
tionship for ion cyclotron waves 



--(■-P'J)- 



The quantitative divergences (stronger experimental dependence of resonance 
frequency on plasma density) do not as yet yield to a satisfactory explana- 
tion. 

REFERENCES 

1. Hooke, W. M. , Rothman, M. A. Nucl. Fusion, 4, 33, 1964. 

2. Shapiro, V. D. In the Book: Plasma Physics and Problems of Controlled 

Thermonuclear Synthesis (Fizika plazmy i problemy upravlyayemogo 
termoyadernogo sinteza) , 1. Izdatel'stvo AN USSR, Kiev, 62, 1962. 

3. Kondratenko, A. N. In the Book: Plasma Physics and Problems of Con- /185 

trolled Thermonuclear Synthesis (Fizika plazmy i problemy upravlyaye- 
mogo termoyadernogo sinteza), 3. Izdatel'stvo AN USSR, Kiev, 91, 
1963. 

4. Kondratenko, A. N. Atomnaya Energiya, 16, 399, 1964. 

5. Faynberg, Ya. B. Atomnaya Energiya, 6, 431, 447, 1959. 

6. Faynberg, Ya. B, In the Book: Plasma Physics and Problems of Con- 

trolled Thermonuclear Synthesis (Fizika plazmy i problemy 

178 



upravlyayemogo termoyadernogo slnteza), 1. Izdatel'stvo AN USSR, 
Kiev, 20, 1962. 

7. Kornilov, Ye. A., Kovpik, 0. F. , Faynberg, Ya. B., Kharchenko, I. F. 

In the Book: Interaction of Charged Particle Bundles with a 
Plasma (Vzaimodeystviye puchkov zaryazhennykh chastits s plazmoy) . 
"Naukova Dumka", Kiev, 23, 1965. 

8. Orayevskiy, V. N. , Sagdeyev, R. Z. Zhurnal Teoreticheskoy Fiziki, 

32, 1291, 1962. 

9. Kondratenko, A. N. Zhurnal Teoreticheskoy Fiziki, 33, 1397, 1963. 

10. Kino, G. S., Ludovici, B. Proc. IV Intern. Conf . , Uppsala, 2, 762, 

1960. 

11. Mints, A. L. , Rubchinskiy, S. M. , Veysbeyn, M. M. , Vasil'yev, A. A. 

Radiotekhnika i Elektronika, 1, 974, 1956. 

12. Ivanov, B. I. In the Book: Plasma Physics and Problems of Controlled 

Thermonuclear Synthesis (Fizika plazmy i problemy upravlyayemogo 
termoyadernogo sinteza) , 3. Izdatel'stvo AN SSSR, Kiev, 54, 1963. 



179 



SECTION IV 
EXCITATION OF PLASMA OSCILLATIONS 



RADIATION OF ELECTRONS IN THE PLASMA-MAGNETIC FIELD /186 

BOUNDARY LAYER 

V. V. Dolgopolov, V. I. Pakhomov, K. N. Stepanov 

The cyclotron radiation of electrons in the plasma-magnetic field 
boundary layer can make a significant contribution to the energy balance 
of thermonuclear reactors with a small density, which employ magnetic 
grids to contain the plasma. This problem was examined in (Ref. 1). 

The thickness of the transitional layer between the plasma and the 
magnetic field may comprise several Larmor electron radii 



Pe = — [Ve = }/ -, tOfi = — 5 . 
"*bV y tn' ° mc j 



The trajectory curvature of electrons having a velocity on the order of 
Vg is also on the order of p^. Moving along such a trajectory in a 

vacuum, a non-relativistic electron radiates the following energy per /187 
unit of time 

S'-#^^ ^«-^ (1) 

The number of emissive electrons per unit of layer area is npPg (n. — 
plasma density). Therefore, the total intensity of cyclotron radiation 
from unit of layer surface — if it is assumed that all the electrons in 
the layer radiate the same way as in a vacuimi — is 

3 (2) 

, dw e^/Zo^flt/J 



dt 



«oP«- 



Since 



Bo — 8tz n^T, ^2) 

^-""^V^' (4) 

where a '\' 1. Expression (4) coincides, within an accuracy of a coefficient 
on the order of unity, with the result derived by Burhardt (Ref. 1). 



180 



We may employ expression (1) only if the refractive index n for 
radiated frequencies u '\' uig is close to vinity. In the case under con- 
sideration, the radiated frequencies lie in the region of the anomalous 
skin-effect 



Re/z~Imn~^~f »1 (5) 



(. = /: 



ine^n, 







A 



— plasma frequency). In a medium with a large, complex 

refractive index (5) the radiation intensity differs greatly from the 
radiation intensity (1). For example, in a dense, non-relativistic 

plasma I Q2^ — o)|) » the intensity of cyclotron radiation of the main fre- 
quency decreases, as compared with radiation in a vacutmi, by a factor of 
^ (Ref. 2-4). 

Let us determine the intensity of cyclotron radiation, assuming that 

the radiated waves correspond to stable plasma oscillations. In the case 

of radiation equilibrium, the radiation flux falling on the plasma-magnetic 

0) T 
field boundary, Ipj = ,32 » equals the sum of the flux emanating from 

the plasma I(w)and the flux reflected from the plasma IrjR (R — reflec- 
tion coefficient). We thus find that I((jj)= Irj(1 - R) • In the case under /188 

consideration, 1 - R '\^ — '\' — ^. The width of the radiation spectrtmi, 

n c 

caused by the Doppler effect during radiation and by the nonuniform! ty of 

the magnetic field in the boundary layer, is on the order of a>g in the 

case under consideration. Therefore, the total intensity of electron 

cyclotron radiation in the layer is 



/ ~ / ((b) COa . 



i^%v^T e^n'gf'T^ (6) 



A comparison of (6) and (4) shows that allowance for plasma polariza- 
tion decreases the intensity of cyclotron radiation by a factor of 



2 
mc 



Due to plasma resonance, radiation of boundary layer electrons strongly 
increases in the region of frequencies o) which are less than the maximimi 
Langmuir frequency fig* ^^^ considerably greater than ojg. In the resonance 



region where Q (;t) = "l/ "^"^'"o W „,^ the intensity of electron braking 

radiation greatly increases at the frequency o). 

Let us investigate absorption of waves whose electric vector lies in 

181 



the plane of Incidence (the XY plane) , and who Impact on the plasma 
layer from a vacuum. The electric field component which is parallel 

to the plasma boundary, Ey "^ gikyy-iw ^ satisfies the following equa- 
tion 

^^ 7 P^'S- IT + i?^ — '^>'j^J'="' (7) 

where e = l gJM ■ • ^ (x) Q' (x) ±s the dielectric constant of the plasma; 

V (x) = - " f_MjO^ . — frequency of collisions; A — Coulomb logarithm (it 
VmT^x) 

c c 

is assumed that v << to). Since Q. '^ — wb, the wavelength X = — is on the 

order of the layer thickness. 

At the point where Re e = (region of plasma resonance) , the wave 

electric field increases sharply (£„ '^ E^ In e -^ Eq In — ; Ex '^^ — 2. "^ — Eq , /189 

Eg — amplitude of incident wave) . This leads to the fact that a consider- 
able portion of the incident wave energy is absorbed in the layer having 

the thickness Ax ^ X—, in the vicinity of the resonance point. Therefore, 

the intensity of braking radiation at the frequencies w < fi is 

r r o ^ '""'o'^ (8) 



c^m'^' 



2 
mc 



Consequently, the intensity of cyclotron radiation is — ;j;— times less than 

the braking radiation intensity. 

2/3 
We may employ equation (7) only in the case ofv>vo = B f2. 

If V < vq, we must take into account the formation of plasma waves in the 
resonance region. 



REFERENCES 

1. Burhardt, H. Nucl. Fus., 2, 1, 1962. 

2. Ginzburg, V. L., Zheleznyakov, V. V. Izvestlya Vuzov. Radiofizika, 

1, 2, 59, 1958. 

3. Stepanov, K. N. , Pakhomov, V. I. Zhurnal Eksperimental'noy i 

Teoreticheskoy Fiziki, 38, 1564, 1960. 



182 



4. Pakhomov, V. I., Stepanov, K. N. Zhurnal Teoreticheskoy Fizikl, 
33, 43, 1963. 



RADIATION OF LOW FREQUENCY WAVES BY IONS AND ELECTRONS 
OF A NON-ISOTHERMIC MAGNETOACTIVE PLASMA 

V. I. Pakhomov 

As Is knovm, the propagation of three normal waves — Alfvdn, rapid 
and slow magnetosound waves — is possible In a strongly non-lsothermlc 
(Tg << T^) magnetoactive plasma, In the low frequency region (o) ^ ^^i) • 

Each of these waves may be excited due to Ion cyclotron radiation or due 
to Cherenkov radiation of Ions and electrons moving along a spiral In 
this plasma. 

This article determines the expressions for the Intensities of /190 
radiation of these wave types by Ions and electrons. The emissive and 
absorbant capacity of the plasma are determined In the frequency region 
0) ^ '^x" ■^^ case of a low-pressure plasma is investigated in detail, 

H 
when the Alfvdn velocity v^ = yr rr considerably exceeds the speed of 

sound in the plasma Vg i/-J^ • In this case, the refractive index of a 

^M 
slow magnetosound wave is considerably greater than the refractive indi- 
ces of the two other waves. Therefore, it is natural to expect a sharp 
increase in the radiation intensity of a slow magnetosound wave. It is 
shown that in a low-pressure plasma the intensity of cyclotron radiation 

. 2S+1 
of a slow magnetosound wave by an ion for the s-th harmonic is 



(^) 



times greater than the radiation Intensity of Alfvdn waves and rapid 
magnetosound waves . 

Cherenkov radiation of low frequency waves by electrons of a non- 
isothermic plasma may make a basic contribution to the over-all plasma 
radiation. It is shown that the ratio of the intensities of ion cyclotron 
radiation and Cherenkov electron radiation in the case of u) ffe ^"^1 ^^ °^ 



183 



the order 



-©"1"^)- - 



ere Vrp- and v™, are the mean thermal 



velocities, respectively, of ions and electrons, and v» is the phase 
velocity.* 



Propagation of Electromagnetic Waves in a 
Non-Isothermic , MagnetoactiveP lasma 



The general expressions which are given in the appendix in (Ref. 2) 
may be employed to derive the dielectric constant tensor of a non-iso- 
thermic plasma located in the outer magnetic field. The svmi with respect 
to particle types — i.e., electrons and ions — must be taken. Let us 
assume that the following condition is fulfilled 



(Aa = 



,— to 



"j."tci 
"Ha 



<^\, a. =i, e; 



2„ = rr^. 



y-i k , v^ 



>1. / = 0, ±1; 



■ h 



^le = r^= 



■fie 



y2k,V. 



»1, /=±1, ±2, 



I "re 



2oc = . 



«1. 



These conditions are fulfilled if it is asstmied that the plasma is greatly /191 
non- is o thermic (T^ >> T^) . As a result, for the tensor of the plasma di- 
electric constant we obtain the following expression 



where 



e<7 (k, w) = a8y -f chihj + dtijkhk + 
+ e (x,- [xA]/ — r., [xA],) + / [xA]i [f.h]j. 






u—\ ' 



c = 



1 — u 



+ v 



n^ cos^0 



q{Zoe) — l 



d = - 1 ^ [j4r^ + ^ (20.) tg^ 9 ] sgn . 



e^-iJS°'^ 



y^u cos'' Q 



sgn (d; f — iy 2% 



Mu cos 6 ' 



(1) 



* A portion of the results have been published in (Ref. 1) 



184 



where 



g {zoe) = 1+1 K'^zo.; P« = -f- = K ;;i?'' • 



e is the angle between the direction of wave propagation k and the 
direction of the outer magnetic field H. If the frequency u) is close 
to the harmonics of the ion gyrofrequency stog^ (s = 2, 3, ...), the 

following expressions must be added to (1) 

a' =—c' =^id' =i=2ia„ 

where 



n vs^ (s^n s i n G)^"-^ -2^ 



2's! pn I cos e I 

(2) 



0> SO) 



Zs = 



//« 



K2 ^y^^ cos e 



■^ c y Mel- 



The dispersion equation determining the longitudinal refractive /192 
index nn = n cos 6 as the function of the transverse refractive index 

n_|_ = n sin 9 and the frequency w has the following form in the given 

case [see also (Ref. 3)] 

6 . 4 /" — 2 2 2 2u \ 2 r 9 "^1 + 2o , „ 

ni +n,(^—^n,-n,-^^n][nl-^^ + nUf~ (3) 

~~M+ 2n!)] + ^ [nl {n! + n^) -n!n%] = iA, 



where 



+ u~\ + —Til— 1«. -- i7iri-/l + 2K' \n . [-V + V + 

■ „2 [(H - T^)" + 2 2 9„2,«*l1 2/2, 2 A 



185 



When there is no damping (A = 0) , equation (3) has three solutions 
1^11 = 1^ IP (J = 1» 2, 3). Assuming that 

«! =n,i + inii, n\j<^n,j, 
we obtain the following expression for ri\\^ 

A {"■ I /) (4) 

The damping coefficient of the j-th normal wave is 

"/^y^i/cosx. (5) 

where x is the angle between the directions of the outer magnetic field 
and the energy flux of the corresponding wave. 

The dispersion equation (3) may be solved in two limiting cases : If 
the Alfven velocity is considerably greater than the speed of sound in 
the plasma (va >> v ) and if the frequency co is considerably less than /193 

the ion gyrof requency (u << w^) . 

Let us examine the case when v. >> v . The dispersion equation (3) 

A S 

assumes the following form 

„« 4- A t-^f ni -ni)^n\ [n\ (^ + .;) - („ 

We may find one of the solutions for equation (6) by assuming that n| '^ 
nj_ "^ ng >> n^. As a result, we obtain (Ref. 4) 



n n 1 = «s + 



2 ,2 , "JL (7) 



The imaginary part of the refractive index of a slow magnetosound wave (7) 



xs 



«!! 



« I l"x 



tW'-t^'')- <^> 



The first component in the right part of equation (8) takes into account 
Cherenkov absorption of a slow magnetosound wave in an electron gas. The 
second component takes into account cyclotron absorption in an ion gas . 

We may find two other solutions for equation (6) , corresponding to 

186 



Alfv^n waves and rapid magnetosound waves, assuming that nn '^' nj_ '^ 

'^ n^ « ng. The radiation of waves corresponding to these two solutions 

was studied in (Ref. 5, 6). 

In another limiting case, when the frequency o) is considerably less 
than the ion gyrofrequency u^x (region of strictly magnetohydrodynamic 

waves) , the dispersion equation (3) assumes the following form 



n, —riA) [n, —n, [ris + rtA —n^LJ—riAnj. + 
+ nl {tiA — ri]_)] =iA', 
where 

A' = -|/| • ^.[in', -n\) [nfin] ~n\ + nl) + 
+ 2n\ hWa] + ^ [2n\ n\ [n] — n\—n\)— n\ {2n\ — n\ + n3i) ]) . 



(9) 



(10) 



The solution of equation (9) /194 

ni,=«^+o(-^) (11) 

corresponds to an Alfvdn wave. The imaginary part of the refractive index 
of an Alfv^n wave is 



"'.-/: 



-^n\(r^, + n\f (12) 



n/n 



Two other solutions (Ref. 7) of the dispersion equation (9) 

n%.3 = \[nA+nl — n\ ±]/' [nl— n\^ n\Y + AnWr] (13) 

determine the refractive indices of rapid waves and slow magnetosound waves, 
The imaginary parts of the refractive indices (13) determining the damping 
of magnetosound waves (Ref. 7) are 



Az',2.3= y\ 



wn "s ("^1 2.3 - "^ + n\) + 2/t^, 2.3n^«i 



187 



General Expression f^ox A on Radiation Intensity in a 
Non-Isothermic Plasma 

Let us employ the general expressions (Ref. 4, 6) for determining 
the strengths of the electric and magnetic fields produced by an ion 
moving along a spiral in a strongly non- is o thermic magnetoactive plasma. 
As a result, the components of the electromagnetic field produced by the 
ion in the wave guide zone assume the following form 



where 



El,^s=PiSlnW,iE^J; 



(i) . 



H 



£«=/'/ cos ^,y£l']; 
//« = ±Pini cos W,iH% 



(15) 



/195 



Pi = 



2ea,; 




"H^ V'' 



sin X 



cos X • 



d^n 



rf 



dn^. 



"K/-«'OW/-"U)' 



^s/ ={±k^sin')(. + ktiCosx)R—'o,jt + s^ + ^ — sf + 



+ -^sgn(cosZ 






£"?] = X («i£ii + n 1 1^33 — 611633) ^s — U-v («x — n i /«x — 

L«x'o 



• 633) '6i2 — V 1 eiife23 + Vinii-rij^ {isj^z — iHz ctg G) 



Js-, 



+ 



£■2 = ± COS X I y ^ [rti (/e^a — ^£23 ctg 6) — fsijEgaj/s + 
7-(ny«i— n/633— "i^ii + 6iies3 + S23) +Uiirti//ZxX 



X 

X («/ — En) + u J 61262 



/s| — sin X I f X [« 1 /"x (J6i2 



^Kr, 



IT (/J/^ 1 /1x — « 11 /«x6ll + 612523-) — 



- £623 ctg 6) + 6112623] "^5 + 

— V i (rt/rt^ii / — «/6n — n^i /611 + Si2e23 ctg e)j /jj ; 

Hf^ = —Vi, cos 6 (f£i2633 + eiiiS23 tg 9) A — 

— ~- [tijess — £11633 — £23 + £11623 tg 6) cos 9 + y B E^n • sin g] J 

l-«x''o J 



(16) 



In expressions (16), v_|_ and vn are, respectively, the ion velocity 
components which are perpendicular and parallel to the direction of the 



188 



outer magnetic field; rn = -— — Larmor radius of aa ion; the argument of the 

<^i 
Bessel function Jg and their derivatives J'g equals '^jj^o- The radiated 

frequencies are 



u>s/ = SMffj -i^k H (U>s;) V 5 . 



(17) 



In the case of s = (Cherenkov radiation) , equation (17) assumes the 
following form. 

(18) 

In expressions (16) nj_ and kj_ are the positive solutions of the equa- /igg 
tion for the saddle point 



^'^»/(«x)_<^^tl/(^x) _, 

= + tgx- 



da,. 



dk. 



(19) 



The negative solutions correspond to saddle points which do not lie on 
the integration path, and therefore cannot make a contribution to the 
integral. The upper sign in equation (19) , as well as in (16) , corres- 
ponds to a cylindrically diverging wave; the lower sign corresponds to 
a cylindrically converging wave. When there are several positive solu- - 
tions for equation (19) , it is necessary to take the sum of all the solu- 
tions. Each solution for equation (19) determines the relationship between 
the angles x ^^'^ ^> i.e., the wave toward which the phase velocity is 
directed at the angle 8, and the group velocity at the angle x to the 
direction of the outer magnetic field. 



We may find the intensity of cyclotron (s i= 0) and Cherenkov radia- 
of an 
expression 



tion of an ion per unit of solid angle Wg.: by employing the following 



where the bar designates averaging over time. As a result, we obtain 



w^; = — a 



sinX 



'^f^APsj^''^^''' 



cosX- 



^'^■(^-^^^)hn«^.-«1.}V.;-4.)^ 



(20) 



where 



^./ = £^]5^] + cos(z + 8)fef. 



189 



In the case of s = 2, 3, 



formula (20) is valid for the cases 



of slow (vp Vrj,^) and rapid (v|| » Vrp^) ions. In order to determine the 

radiation intensity at the main harmonics (s = 1), expression (20) may- 
be only employed in the case of rapp.d ions. In the case of s = 1, the 
results given in (Ref. 4) must be employed to determine the radiation 
intensity of slow ions. For Cherenkov radiation (s = 0) , formula (20) 
is also valid only in the case of rapid ions. Cherenkov radiation of 
slow ions is greatly absorbed. 

Expression (20) for ions with a velocity on the order of the mean 
thermal velocity of plasma ions has the following form (in the case of 



Ws, 



where 



Us: 



\2s— 2 



s-^{snJ^Y 



«/«x«i; 



2'=^(s!)2sinx 



d% 



cosx ■ 



11/ 



dn^ 



-'{Ai-<i)\Ai-'^\kf 



X 



X I [ + n? sin (x: + 6)(n/n j. — s^ sin 6 + isja sin 8 — h^s cos 6> + (s^^ — 
— iHi) (S33 cos X qr U^.^ sin 1) + efg cos X — SggraZ cos X ] cos 8 [{s^i — ^ 
— i^ 12KH3 — ^623 ctg 6) — nfsss + efg] + cos (X. ^r 6) [(s^i — h^^) x 
X (S33 — n^) — mn s / (633 cos 9 + is.^g sin e)]^}. 



(21) 



/197 



Averaging expression (21) ovex the ion distribution, which is 
assumed to be a Maxwell distribution, we inay determine the contribution 
made by ion cyclotron radiation to' the emissive power of the plasma 
at the 0) R; sa3jj£ frequencies 



f}s} 






(22) 



(2iif''cii I / ?■ 



^i 92s— 2 ' 



where uq is the plasma density. 



Cyclotron Radiation of a Slow Magnetosound Wave in a 
Low-Pr4ssure Plasma 

It was noted above that the dispersion equation (3) may be solved 
if the magnetic pressure p-r, is considerably greater than the gasokinetic 

pressure of electrons p^ = n^Tg (|:his is equivalent to the condition 

^A ^^ ^s) • ^^^ ^^ investigate th^ cyclotron radiation of a slow 



190 



magnetos ound wave corresponding to the solution (7) of the dispersion 
equation. 

Employing the general expression (20) for the ion radiation inten- 
sity and taking the condition n^ « n into account for the intensity 

of the radiation of a slow magnetosound wave by an ion at the s-th har- 
monics per unit of solid angle, we obtain the following expression 






-2xjii? 



(23) 



where 



U,i = 



n\ sin .6 



o^n^ J sin X 



»-'^^(-'.^) 



fn?sin(x — G)[r;xsin9x 



X 



X (rei2 — fs23 ctg 0) /s + (i^ sin 9 — a „ cos 6 j nf/,- o^ (%3f£i2 + ' 
£111623 tg 6) /s -f \~r S33 cos 6 -^ u', Btj_ sin 6 j nf/^ -f cos (X — 6) x 
X [f X (eii4 + 633"^ 1) /; — (ZS12 — /S23 ctg 6) (^ tg 6 -f o , ) X 

Kjirex/J I. 



In the case under consideration, the refractive index is determined 
by formula (7) , and the damping coefficient is 



Xsl 






(24) 



7198 



For ions whose velocity is on the order of the mean thermal velocity 
of plasma ions, expression (23) may be simplified (in the case of s = 2, 
3, ...): 



»2,.,2r2 



Wsl = ■ 



e'<xK,, _2. 



2nc 



Usie 



(25) 



where 



U. 



si 



s2 (s^^_n^f^~^tl\n\ sin (X— 6) 



22^(s!)?o(k— l)rt^,lSinX 



cosX- 






s 1 



191 



In this case, the equation for the jsaddle point (19) yields the following 
dependence between the angles x ^^^ ^' 



tge = («-i)tgx. 



(26) 



The contribution resulting frqm magneto-braking radiation of a slow 
magnetosound wave by ions to the emissive power of the plasma, for fre- 
quencies close to swjjj^, is 



— ^Hi?o^(^P"x)^'~'"i "fsi"(fl — y-) -z"^ 
'" ~ (2jij'/=2's! vcn^n ^ , sin X | cos X| ■ ^ '" 



I 
In order of magnitude, ^s, --^ w^riA., 



2s— 2 



(27) 



~l , where wq. is the total /199 



intensity of ion radiation in a vacutim. It was shown in (Ref . 6) that 
the intensities of cyclotron radiatiion of an Alfven wave and of a rapid 
magnetosound wave (j =2, 3) are 



Ws2, 3 — teio«A 



2s— 2 



Comparing these results, we find that 



"si 



"52,3 



■t;^\2s+l ^ 



Thus, in a greatly non-isothermic plasma having a low density, the 
intensity of cyclotron radiation of a slow magnetosound wave for the 

fVA2s-fl 

s-th harmonics is, in order of magnitude, [~j times greater than the 

radiation intensity of an Alfven wa;ve and a rapid magnetosound wave. 



Cherenkov Radiation of Magnetohydrodynamic Waves by Ions 

In the case of u « (jJjj^, the dispersion equation (3) has three solu- 
tions corresponding to magnetohydrodynamic waves . The refractive indices 
of these waves are on the order of jng (or n.), and are determined by ex- 
pressions (11) , (12) . It follows from the condition of Cherenkov radia- 
tion 3|] n||. =1 that frequencies oj << tog^ may exist during radiation 

of rapid ions, whose velocity is 



N 



p/r 



192 



Let us investigate Cherenkov radiation of Ions in the low-frequency 
region o) << okj- (u >> 1) . In order to determine the Cherenkov radiation 

Intensity, let us employ the general expressions (20), assuming that 
s = and u >> 1. In addition, let us assume that the arguments of the 
Bessel function are small 



k^ro = -p « 1. 



(28) 



Taking into account (28) for the Cherenkov radiation intensity of 
an Alfvdn wave, we obtain the following 



Woi = 



8to, 



u^ cos 6 sin X 



n?sin3 6[4pi(«f-«^)+^?] 

j[ _ 

2 d> 



d^n ,1 , dn , , 



dn\ 



-e-2'oi«. 



{n\~n\,f{n\-n\,) 



(29) 



/200 



The damping coefficient of an Alfven wave is 



"oi -if -Km 
*°l=V^°^^K 8A?- 






(30) 



un\n\ 



Equation (19) for the saddle point assumes the following form in this case 



n\ - — ' n\ tg X + n\ (nf - n\) = 0. 



(31) 



It follows from equation (30) that the Alfvdn wave is radiated within the 
limits of a narrow cone ( X '^ ~ ) along the direction of the outer magnetic 
field. 



The Cherenkov radiation Intensity of rapid and slow magnetosound 
waves per unit of solid angle, when condition (28) is fulfilled, is 



i^02.3 



^ ^°'"02.3 _ A 2.3 s,-n3 6 co^(Xj:J) \f^{n\ 2,3 -<)+2 ]^ _2,^^_ 



Stici 



u cos^ sin X 



cosX 



d^n, 



dn 



2t3 D 2,3 1/2 2 \2 



3«. 



(32) 



The damping coefficients of magnetosound waves are determined by 
the expression 



v.02.3 = -^ cosx K 87i? • ■ 



^02. 3 . 
C 



V\ 



"X2,3K2,3-«|3,2) 



(33) 



193 



dn |u 2 2 
The figure presents graphs of the fvmctions — — ^ and n||j(nj") for 

the case nn > and n^ > ng. The points at which the curve "J inter- 

dni 

sects the line y = + tgx correspond to the solution of equation (19) . /201 
As may be seen from the figure, a slow magnetosound wave (j = 2) repre- 
sents a cylindrically converging wave. There are thus two waves for 
each angle x *^ Xmax* ^^ ^^^ case of x > Xmax' ^ slow magnetosound wave 
is not radiated. The angle x = X is determined by the following 
equations 

The rapid magnetosound wave (j = 3) represents a cylindrically 
diverging wave. For any angle x> °^^ such wave is radiated. Expres- 
sion (32) for a slow magnetosound wave is not valid in the limiting 
case, when nj_ ->- o° and n?| -y n? + n^. In this case, we must employ the 

general expression (20) for u >> 1 and s = 0. (We should point out that 
Cherenkov radiation in a direction which is perpendicular to the direc- 
tion of the outer magnetic field may be absent since the condition 
B II n|| = 1 thus contradicts the assumption m « a^. .) If n. < Ug, then 

the substitution n^ J iXq must be made in the figure. In the case of 

n| < 0, the functions ^Hj change their sign. 

dn^ 



Electron Cherenkov Radiation 

It is of interest to study Cherenkov radiation of low frequency 
waves (cj '^ '^±) by electrons of a non-isothermic plasma, because under 

certain conditions it may make the main contribution to the over-all 
plasma radiation. In particular, this is related to the fact that the 
intensity of ion cyclotron radiation sharply decreases as one recedes 
from the center of the line o) = saijj. (s = 1, 2, ...). 

We obtained the expressions for the intensity of Cherenkov radia- /202 
tion, by an electron moving in the plasma along a spiral, of each of 
three normal waves. We determined the contribution made by Cherenkov 
electron radiation to the emissive power of a non-isothermic plasma at 
the (0 ^ ojjj-j^ frequencies. 

We shall employ the general expressions given in (Ref. 2) for s = 
in order to determine the components of an electromagnetic field 

194 



s •« 








Figure 1 



produced by an electron In the wave zone. As a result, we obtain 

£,,=P,sin('F + a|-)£^.: 

£x/=P;COs(w + a^)^^,; 

//,/= ±P/n,cos(w + a^)^^,;, 
Hxj = — tij cos (x + 6) £■?/. 



where 



a = sgn^cosx-^j. 
£»,• == y J. (rtieii + n^, ,£33 — 611633) /o — y 1 [" I /ix ('612 — 1623 ctg 8) -h 

£](/== ±COS^ (yx[«i(fei2 — 2e23ctg6) + ieiaSajj/o — t),[rt,/rex(n/ — 

— i^ii) + Hi^23\ h] — sin X {ux [« II /"x ('612 — '?23 ctg 9) + ^iih^aVo + 

+ , [tifnu - nUii (1 + cos== 6) + ef, + s^zj/oh 

//^,- = — yiC0s8(tSi2e33 + Eii'S23 tg 9) ^0 +v,s\nt[eii{nj — 

— ^ii) + 2i2e23 ctg 9 — efj] Jo- 



(34) 



(35) 



In (34), V|| and vj_ are, respectively, the electron velocity components 
which are parallel and perpendicular to the direction of the outer 



195 



magnetic field. The argument of the Bessel functions and their deriva- 

vi 
tives equals kj_ rg, where rg = ■ is the Larmor electron radius. The 



= ZL 

■^e 



remaining notation was presented above. 

Employing formulas (34) and (35) , we may obtain the following ex- 
pression for the intensity of electron Cherenkov radiation per unit of 
solid angle 



nfiR'P^, 



Wj = 



SltMp, 



\dn, 



I] 



[£;t/^W + cos(x+6)fe/)1. 



(36) 



da> 



The argtmient of the Bessel function is small for electrons having 
a velocity v_l which is on the order of the mean thermal velocity of 



plasma electrons (vj_ '^ V'pg = 

k±ro 




e"): 



m 



«I. 



In this case, expression (36) may be simplified 



2x„ . R 



Wi = 



2m , \ d^n^j 

0(1 — u) sin X cos X — 2~ 



dh 



'If 



d(o 



{A-n\:f{n\,-n\,Y 



(37) 



where 






/203 



Averaging (37) over the electron distribution, which is assumed to 
be a Maxwell distribution, we may find the contribution made by electron 
Cherenkov radiation to the plasma emissive power at the frequencies 

0) 'V av,. : 



'^\ 



_ e==<flrto 




";ig 


(2't)'/'t;„ 


o(l — «)sinX 


cosX ^^,^ 



{Ai—\^\-\i—\.Y 



(38) 



196 



Comparing the contribution made. In order of magnitude, by ion cyclotron 
radiation (22) and by electron Cherenkov radiation to the plasma emissive 
power, we obtain 



1, W; V*}' 



(39) 



It can be seen from (39) that the relative contribution of the elec- /204 
tron Cherenkov radiation increases with an increase in the harmonic 
number s, since It was assumed from the beginning that v^^ << v$ << v,j,g. 

Let us investigate the Cherenkov radiation of a slow magnetosound 
wave by electrons in a low-pressure plasma (ng >> n^;^) . We shall assume 

that the argimient of the Bessel functions in (35) is small (kj_ rg << 1) . 

This can be fulfilled only for very rapid electrons (vj_ >> v-pg) . Employing 

the general expressions for the electromagnetic field components (34) , 
(35) , the equations (7) , (26) , and also the condition of electron Cheren- 
kov radiation, we finally obtain 

* 2«t<, • tg*X{sec'X-f,nlY ' p^ n^n^ | cosX| ' 

, Expression (40) determines the Intensity of Cherenkov radiation for 
an electron having the velocity 3 1| per unit of solid angle in the direc- 
tion X- If 3|| ng < 1, then the radiated wave represents a cyllndrically 

converging wave (for x < — ) . The waves may thus be radiated in any direc- 

2 

tion X. The frequency of a wave radiated at the angle x is 

(41) 



y\sec'X~f,n^X\ ■ 

In another limiting case, when 3 || Ug > 1, the radiated wave repre- 
sents a cyllndrically diverging wave. The vector of the group velocity 
is thus directed at the angle x > Xmin' where 

^min=^'"'=s^(P'"»)- 
The phase velocity in this case is directed at the angle < 0_,g,. 
in the direction of the outer magnetic field, while Qjjgj, = Xmi n • However, 
we cannot employ the expressions obtained in the case of x "*■ Xmln' since 
in this case m ->■ <", according to (41). 



197 



Averaging (40) over the electron distribution, we obtain the 
contribution produced by Cherenkov radiation, by plasma electrons, of 
a slow magnetosound wave to the plasma emissive power: 

We should recall that at any angle x > Xmiji' electromagnetic waves with /205 

two frequencies are radiated — one with the frequency o) > tOg- (electrons 

with 3 11 Ug > 1) and the other with the frequency o) < uiyij^ (electrons with 

the velocity V|| < v ). Expressions have been obtained (Ref. 6), in which 

it must be assumed that Zq^ « 1, for the intensity of Cherenkov radiation 

by electrons of an Alfv^n wave and a rapid magnetosound wave in a non- 
isothermic plasma with a low density. 

REFERENCES 

1. Pakhomov, V. I. Zhurnal Teoreticheskoy Fiziki, 34, 16, 1964. 

2. Pakhomov, V. I., Aleksin, V. F. , Stepanov, K. N. In the Book: Plasma 

Physics and Problems of Controlled Thermonuclear Sjmthesis (Fizika 
plazmy i problemy upravlyayemogo termoyadernogo sinteza) , 2. Izdatel'- 
stvo AN USSR, Kiev, 40, 1963. 

3. Kovner, M. S. Izvestiya Vuzov. Radiofizika, 4, 765, 1035, 1961. 

4. Stepanov, K. N. Zhurnal Eksperlmental'noy i Teoreticheskoy Fiziki, 

35, 1155, 1958. 

5. Pistunovich, V. I., Shafranov, V. D. Yademyy Sintez, 1, 189, 1961. 

6. Pakhomov, V. I., Stepanov, K. N. Zhurnal Eksperlmental'noy i 

Teoreticheskoy Fiziki, 43, 2153, 1962. 

7. Stepanov, K. N. Uspekhi Fiziki Zhidkosti, 4, 678, 1959. 



198 



BRAKING OF RELATIVISTIC PARTICLES IN LOW ATMOSPHERIC LAYERS 

V. B. Krasovitskiy, V. I. Kurllko 

As is known, V7hen a relativlstlc charged particle moves in an external 
field, the braking force by radiation increases proportionally to the 
square of the particle energy, and may be greater than the Lorentz force 
(Ref . 1) . The influence of the medium on the braking radiation is of 
great interest, since — when particles move in the lower atmospheric 
layers — the braking force by radiation may differ significantly from 
the vacuum force. The expression for the braking force by radiation of 
an oscillator in a medium was obtained and studied in the non-relativis- 
tic approximation in (Ref. 2, 3). The spectral radiation density was 
studied in the relativistic case in (Ref. 4-7). 

In order to investigate the influence of the mediimi on the braking 
radiation, it is necessary to know the total energy losses of the particle, 
i.e., it is necessary to specify the medium properties. We shall formulate 
a model for a medium with an isotropic dielectric with a given dispersion /206 

e(«,)=l ^ 

4iTe2no 

where Q are the atom resonance frequencies; to^ = ; tlq — density 

m 
of the medium particles. 

The following expression was obtained in (Ref. 4-7) for the spec- 
tral density of particle radiation in the frequency region w >> Wtt (co„ = 

'^ ^^l - B2 ) 

^("^ ^ -iTTf- '-?(' -pi){2/C|(^)- I^^(^)rf^j.P^e < 1; 
Z' (">) = ^[l - -p^jjl + 2 iy_v. (^) - J.,, ($)] _ f [y_./, (^) _ 



(1) 



whe 



re £=-?.. Ji(i B^eV/i; j m and Kp(g) are the Bessel functions. 



Let us employ formula (1) to determine the total losses by radia- 
tion as a function of the particle energy. Since it is difficult to 



199 



integrate equation (1) in the general case, let us find the frequency 
regions in which the spectral density has the maxima (K "^ 1) , and let 
us determine the radiation intensity in each of these regions. 

1. In the low-energy region, when the following inequality is ful- 
filled 



y < 5 « ;^ « 1. £ = (1 - p V^., 



(2) 



the frequency u '^^ a)„ (1 - 3^) ^ — which makes the maximum contribution 
m n 

to the vacuum (Ref . 8, 9) — is small as compared with the resonance fre- 
quency ojjjj << Q, In this case, the radiation spectral intensity has 

maxima at the frequencies oij^ and Q. The braking radiation at the fre- 
quencies ojjjj causes losses which coincide with vacuum losses 



,2 



"^B-^jr- £5- (3) 

The maximum close to the frequency determined from condition g^e(a)) ^ 1 
is also caused by the curvature of the particle trajectory. However, in /207 
this case the influence of the medium is significant, as a result of which 
the radiation is considerably greater than in the preceding case 



W^p<{-Jfj'\ (4) 



Thus, the braking radiation is larger (in the energy region under 
consideration) than in a vacuum, and increases with the energy E propor- 
tionally to E h. 

2. The inequality (2) assimies the following form for given character- 
istics of the medium and unchanged magnetic field strength with an increase 
in the particle energy: 

^"«J.«:-I«i. ») 

In this case, the braking radiation maximtmi is located at the frequency 
0) c^ ^ . 1 The corresponding loss is 



"0 



W^ — ."^"^* 1 



f F- (6) 



200 



3. With an increase in the particle energy, the relationship 
between the particle parameters and the meditnn assumes the following 
form 



ii,«#«^k<l. (7) 



Thus , the radiation maximum in a vacuum occurs in the frequency region 
m » Q. The force of the braking radiation depends on the parameter 

y = . i in this case. 

tojj E 

In the case of vi >> 1, the braking radiation intensity in the fre- 
quency region to 'X' oij^ is exponentially small as compared with the corres- 
ponding value in a vacuum. The braking radiation intensity is also small 

(U7^ «?£*")" ^^ *-^^ opposite case (y « 1), the influence of the medium 

on the braking radiation may be disregarded, so that the particle losses 
coincide with vacuum losses. 

Thus , the medium influences the particle braking radiation only 
when the inequality Wq >> ^a>jj is fulfilled. In the region of relatively 

small energies /£^<^^] , the presence of the medium leads to an increase /208 



r«l) 



in the braking radiation, and in the energy region E^ >> — r- it leads to 

a decrease in the braking radiation, as compared with a vacuum. Thus, 
the braking radiation is significantly less than the vacuum radiation in 



the energy region — > £ > ( — ) 



The change in the braking radiation intensity in a medium may be 
explained in physical terms as follows. In the presence of the medium, 
the frequency determining the maximum of the radiation spectral density, 
according to formulas (1) , depends on the medium parameters 



"'*"«^^^=7r^'<(^> = ' 



For small particle energies, the presence of the medium leads to an in- 
crease in the frequency %(£ ('»%i) > 1» aiid the corresponding wavelength 

V7 



decreases): w < tOjjj < fi. Therefore, the braking radiation 



201 



intensity, which is proportional to ((») ) '^j also increases. For large 

particle energies (01^ » U) , the frequency uij^ decreases considerably: 

oj* « J2 << a)jjj (the characteristic wavelength X^ increases, e(a) ) < 1). 

Correspondingly, the braking radiation intensity decreases. The 
Cherenkov radiation, which can be computed from the customary formulas 
in the case under consideration, is larger than the braking radiation 
throughout the entire energy region where the influence of the medium 
must be taken into account. 



REFERENCES 

1. Pomeranchuk, J. Journ. Phys., 1, 65, 1940. 

2. Ginzburg, V. L. , Eydman, V. Ya. Zhumal Eksperimental'noy i 

Teoreticheskoy Fiziki, 36, 1827, 1959. 

3. Ginzburg, V. L. Uspekhi Fizicheskikh Nauk, 69, 537, 1959. 

4. Tsytovich, V. N. Vestnik Moskovskogo Gosudarstvennogo Universiteta, 

Seriya Fiziko-Matematicheskikh i Yestestvennykh Nauk, 11, 27, 
1951. 

5. Sitenko, A. G. Author's Abstract of Candidate's Dissertation (Avtore- 

ferat Kand. Diss.), Khar'kov, 1951. 

6. Kaganov, M. I. Uchenyye Zapiski Khar ' kovskogo Gosudarstvennogo 

Universiteta, 49, Trudy Fizicheskogo Otdeleniya Fiziko-Matemati- 
cheskogo Fakul'teta, 4, 67, 1953. 

7. Vaynshteyn, L. A. Radiotekhnika i Elektronika, 8, 1698, 1963. 

8. Schwinger, J. Phys. Rev., 75, 1912, 1949. 

9. Landau, L. D. and Lifshits, Ye. M. Field Theory (Teoriya polya) . 

Gosudarstvennoye Izdatel'stvo Fizicheskoy i Matematicheskoy 
Literatury (GIFML) , Moscow, 1960. 



202 



EXCITATION OF WAVES IN A CONFINED PLASMA 
BY MODULATED CURRENTS 

A. N. Kondratenko 

The excitation of waves in a plasma by modulated currents has been /209 
studied repeatedly (Ref. 1-3). In all of these studies, it was 
assimied that the plasma was an unconfined, isothermic, and linear 
medium. In this article we shall study the excitation of waves in a 
confined plasma, and we shall attempt to take its nonlinearity into 
account. 

When solving the problem of wave excitation in a plasma by a 

modulated current j , we assume that the current is given, and we shall 

disregard the inverse Influence of the wave on the motion of current 

particles. This is valid as long as the energy losses of the current 

particles at the wave length are negligibly small as compared with the 

energy itself. The electric field strength of the wave is determined 

by the equation LE = j , where L is the differential operator, which is 

generally speaking nonlinear. A solution of this equation leads to the 

following value for the Fourier field components: E = J. If the f re- 
's" 

quency of the current modulation is such that A vanishes (dispersion 
equation), then the field strength becomes large. It Is usually assumed 
that A is limited by particle collisions or Cherenkov absorption of wave 
energy by plasma particles. This is the result of the linear theory, 
and It is valid for insignificant amplitudes of the wave field strength. 
However, in resonance, when A 'v. Q, the field strength greatly increases, 
and the linear theory may be invalid. Even in the case of slight non- 
linearity, the limitation on the amplitude of the wave field strength 
due to nonlinear interaction may be more substantial than the limitation 
imposed by dissipation. 

Let us study the plasma layer which is infinite in two directions 
(y, z) and bounded by two parallel metallic plates in a third direction. 
The distance between these plates is 2a. A modulated current having 
the form of an Infinite layer of thickness 2b, b < a, moves in the plasma 

along the z axis. The plasma is located in a constant magnetic field 
directed along the current, which is so strong that the plasma particle 
motion across the field may be disregarded. 

Let us Investigate two problems. In the first problem, let us 
determine the wave field strength in the linear approximation for an 
arbitrary dispersion law. In the second problem, let us determine the /210 
wave field strength in the nonlinear, but hydrodynamlc approximation. 
We shall asstmie that the nonlinearity is slight. The following quantity 
is a small parameter of the problem 



203 



where e^^ and m^ are, . respectively, the charge and mass of particles of 
type a (a = i — ions, a = e — electrons); Eq — amplitude of wave 

field strength in the linear approximation; o) — modulation frequency; 
v^ — phase velocity of a propagated wave, equalling the velocity of 

current particles; u,,, = '^" ^^Ta — mean thermal velocity of a particles) 

We can write the system of equations describing wave excitation in 
a plasma under the conditions being considered as follows 

x[(3-<)| • T + {3-2".\)|4"] = -4-|: (1) 

1 _ gZ / V. 
where E - z = component of the wave electric field; a = Ift = * 



ightj ; 



B2c2 



4Tre2nr 



(• 



c — speed of light); ? = ojt - ksz; Q^ = — (no — equilibrium 

density of plasma particles which is equal for ions and electrons) ; 

u)' = 0) + ±\)^ (v^ — frequency of collisions between a particles) ; f Qct — 

equilibrium distribution function of a particles noirmalized to unity; 
Uj^ = _2L (V(jj — hydrodynamic velocity of a particles) . 

The derivation of equations (1) and (2) is given in (Ref. 4). In 
these equations, we must set y ~ 1» ^nd we must supplement them with 
the boundary conditions , which in this case may be reduced to setting 
the field strength on the wave guide walls equal to zero: 



E{±a)= 0. 



(3) 



We can present the modulated current in the following form /211 

'ep,V„ 0<{x)<b. 



/=;iWsin?, /! = ( 0. b<ix)<a. ^""^ 



204 



where e — charge; po — density; Vq — velocity of current particles. 

We can obtain the equation of the linear approximation for the 
field from equation (1), setting u^^ = in it. The solution of the 

linear equation can be written as E = R(x) cos 5. We obtain the 
following equation for R(x) 

g-l-A^2/? = -47««aA. (5) 




(5a) 



(6) 



where 

;fe2 =-o(o''l 1+5] 
\ «=, 

Solving equations (5) and (3) together, we find 

R(x)= ^/?„cosa„jc, 

n=0 

where 

Jin = -^ — ,ln — ^ -y-^ sm a„o, 

n 1 

a„ = ^(2n+l); A„ = a^-&i. 



If the velocity of the current particles is close to the mean 
thermal velocity of plasma ions or electrons, the integral in formula 

2 2 

(5a) must be determined numerically, Im kj_ and Re kj_ are equal in order 

of magnitude, damping is large, amplitude of the excited wave field 
strength is small, and the energy losses by the current are insignificant. 
In this case, we can confine ourselves to the linear theoiry. 

2 2 
If Vq >> v-fg ^ or v-fg >> Vq >> V.J1J, then Im kj_ << Re kj^ and at a 

certain frequency of current modulation | A^^j "^ Iva. k_j_ << o^^ may hold. 
The corresponding K^ (we shall designate it by Eq) increases greatly, 

and only the resonance component can remain in the sum (6) : R(x) ^ 
!^ Eg cos Oj^X. 

Since the amplitude of the wave field strength is large, the non- 
linearity of the medium may be significant. When determining the non- 
linear dispersion equation, let us regard the nonlinearity and thermal 
scatter as small independent additions to the linear hydrodjmamic 



205 



2 2 
dispersion equation (A^ = 0, a^^ = Re kj_) . Therefore, the wave damping /212 

caused by kinetic phenomena can only be taken into consideration in 
the third approximation. 

We may find the hydrodynamic velocity of the linear approximation, 
corresponding to A^^ I^ 0, from equation (2) 

«(') = Sa cos oLnX sin ?. (7) 

Let us substitute (7) in equation (1), and let us retain the terms "Xiz . 

Then, representing the field E in the second approximation in the form 
E^ ' = EgRaCx) sin 2^, we obtain the following equation for R2(x) 



-^ + q^R% = h cos^ a^x. 



(8) 






a=l 



Solving equations (8) and (3) together, we find 

i?3=A^_/^._J^.£2iif_^.^2!!f^ (9) 

In the second approximation, the hydrodynamic velocity is 



ui'^ = ~^^R. 



,2 






C0S2S. (10) 



After substituting u^-*^), u(^) in the right part of equation (1) of 

a C( 

the first and second approximation, we obtain the following equation for 
the field E in the third approximation 



-^ + --' [y - ^'^> s , ;r .., r - -43 cos 3^ - 



<-(--+ '\)(l-"x\) 

' Zj tt; 9T2 (-^ — «™ I ^a + -o- — ^ — -^5 — 



(11) 



X cos^ a„x ] cos a^x cos ?, 



where Aa is a certain function of x which is determined by the third 



206 



harmonic with respect to K' 

We can write the solution of equation (11) in the form E^^^ = /213 

= EgRaCx) cos 5. We obtain the following equation for RsCx) 

"S?' + *i ^3 = Qi cos a„x + Q2 cos gx cos anX + Q3 cos 3a„x, (12) 



where 



-4. y 



-V . 3 

n 

9- 8ui 



qi ' 8 ^" ^ 



^4(i-u2J2 9"^ W4a2_92 COS 9a • 

Since Qi 2 3 '^ e^, we can write the solution of equation (12) in 
the following form (Ref. 5) 

/?3 = cos']> + BiCos(a„+ q)x + B2Cos{o.„— 9) a: + B3 cos 3a„A;, 



a=^.+sM. 



We thus find 



cos 



(b -.^\y^Q2 p°sK+g)x cos (a„ - g) X -\ 



-^ cos 3a„jf. 

Employing the boundary condition (3) , we find that 

(O, ^ n 2a„sin(7a x_ „v 

^«xy g ^a^ — cl" 

We can write the solution of equation (13) as follows 



We find „ ^ ^ 4a^Q 



2a„8 = Q,+ j:p^,smga 



(14) 



207 



Thus, we can write the nonlinear hydrodynamic dispersion equation I2\.k 
as follows 



i-'+? .(.+<v.K.- ^]°'.+^-'- (15) 



and V includes both pair collisions and Cherenkov wave absorption by 
plasma particles. 

In the case of resonance, when Re k_|_ = a^, the field strength ampli- 
tude of the excited wave is 



^8 — 



|(^-)-{"ET^.-5/f 



(16) 



Let us first examine high frequency oscillations: Up. g << !• Dis- 
regarding the electron mass as compared with the ion mass , we obtain 

2a„8 = -as^Qg 8,; 8, = | . . ^""'"^^ - ("°--"')'' . !M£ (17) 
' 0.1. X 8 4^._22^ Qg.C-'-Sy "" * 

We can show that in the case of (u ^ ^Oe> 5i > always holds. We obtain 

the following value from the equation (15) for the square of the phase 
velocity 

P Qg, - <o» + c^.l L^"' " + Sg^ - .^ + ^^l J • (18) 

Since 6i > 0, the phase velocity increases with an increase in the field 
strength amplitude of the wave. We may use formula (16) to determine 
the amplitude of the field strength: 

£. ^ S'toiepoV, sin a„ft 

" ^i{(,xr-v\i]Y ""'' • <i9) 



9 *— 

If e 6i > — , the maximum amplitude is determined by the nonlinearity ; 



8^,Vt ^gp. sina„ft (20) 



<8 



"1 



208 



The low frequency oscillations: iijig >> 1 >> "xi* 

The amplitude of the wave field strength in this case is /215 



'" ^.[c*)-+e)"r °"° • 



HS) 



If «?eF^> 1 ((* = — ), then 62 and the phase velocity have the values 



of (17) and (18), respectively, for a replacement of the indices e ->- i. 



q2 

In the case of sound oscillations, when ">*^— ^, we have 






(22) 



Vl = l^v 



^li (1 + ^'»2) (23) 



" ^l + ->'r 



If — ^ < e?6„, the field strength amplitude is limited by the nonllnearity 

and has the same form as in the case of high frequency oscillations (2) , 
with a replacement of the indices e ->• i, 1 -> 2. 

REFERENCES 

1. Kippenhanh, R. , Vries , H. Zs. Naturforsch, 15a, 506, 1960. 

2. Kondratenko, A. N. In the Book: Plasma Physics and Problems of 

Controlled Thermonuclear Synthesis (Fizika plazmy i problemy 
upravlyayemogo termoyademogo sinteza) . Izdatel'stvo AN USSR, 
Kiev, 176, 1963; Uspekhi Fiziki Zhidkosti, 7,371, 1962 

3. Aleksin, V. F. , Stepanov, K. N. Zhurnal Teoreticheskoy Fiziki, 34, 

1210, 1964. 

4. Kondratenko, A. N. Zhurnal Teoreticheskoy Fiziki, 34, 606, 1964. 

5. Bogolyubov, N. N. , Mitropol'skiy, Yu. A. Asymptotic Methods in 

the Theory of Nonlinear Oscillations (Asimptoticheskiye metody v 
teorii nelineynykh kolebaniy) . Fizmatgiz, Moscow, 1963. 

Scientific Translation Service 

4849 Tocaloma Lane 

La Canada, California 

NASA-Langley, 1967 F-4'4-9 " 



"The aeronautical and space activities of the United States shall he 
conducted so as to contribute . . . to the expansion oj human knowl- 
edge of phenomena in the atmosphere and space. The AdministratiotJ 
shall provide for the widest practicable and, "appropriate dissemination 
of information concerning its activities and the results thereof." 

— ^Nationai. Aeronautics and Space Act of 1958 



NASA SCIENTIFIC AND TECHNICAL PUBLICATIONS 

\ 



TECHNICAL REPORTS: Scientific and technical information considered 
important, complete, and a lasting contribution to existing knowledge. 

TECHNICAL NOTES: Information less broad in scope but nevertheless of 
importance as a contribution to existing knowledge. 

TECHNICAL MEMORANDUMS: Information receiving limited distribu- 
tion because of preliminary data, security classificatiqp, or other reasons. 

CONTRACTOR REPORTS: Scientific and technical information generated 
under a NASA contract or grant and considered an important contribution to 
existing knowledge. 

TECHNICAL TRANSLATIONS: Information published in a foreign 
language considered to merit NASA distribution in English. 

SPECIAL PUBLICATIONS: Information derived from or of value to NASA 
activities. Publications include conference proceedings, monographs, data 
compilations, handbooks, sourcebooks, and special bibliographies. 

TECHNOLOGY UTILIZATION'^ PUBLICATIONS: Information on tech- 
nology used by NASA that may be of particular interest in commercial and other 
non-aerospace applications. Publications include Tech Briefs, Technology 
Utilization Reports and Notes, and Technology Surveys. 



Details on the availability of these publications may be obtained from: 

SCIENTIFIC AND TECHNICAL INFORMATION DIVISION 
NATIONAL AERONAUTICS AND SPACE ADMINISTRATION 

Washinston, D.C. 20546